# Code refactoring & technical debt management **Domain:** [Software Engineering](https://www.thestateofplay.ai/domain/software-development) · **Tier:** Bleeding Edge · **Trend:** Steady AI that identifies refactoring opportunities, surfaces technical debt, and suggests prioritised improvements for code quality and maintainability. Includes dead code detection, complexity reduction, and maintenance cost estimation; distinct from code review which evaluates new changes rather than existing code. ## Overview AI-assisted refactoring and technical debt management uses models to find dead code, reduce complexity, prioritise debt and restructure existing codebases, rather than judge new changes. It matters because code structure now carries a direct cost in AI-driven workflows, and AI-generated code is adding debt faster than teams clear it. The practice is a bleeding-edge practice, steady, because named organisations do run it in production. But their successes cluster in narrowly scoped, heavily verified migrations, while the wider deployment record is dominated by incidents, rework and declining refactoring discipline. Verification capacity stays the ceiling until success, not disappointment, becomes the main signal for general refactoring judgement as well as for bounded migrations. ## Current Landscape Deterministic, pattern-constrained refactoring now runs at enterprise scale. Adyen deployed OpenRewrite across hundreds of developers and thousands of modules, generating 4,000 automated MRs in 2 months with a 70% merge rate and a median review turnaround under 2 hours. Google Finance Engineering used Antigravity CLI to migrate 30+ DAOs in a dual-write database migration. Moderne, named a Leader in Gartner's inaugural Magic Quadrant for AI-Augmented Code Modernization Tools, has published a practitioner playbook for getting refactoring merged across hundreds of repositories. It covers platform teams, scoping with finder recipes, custom recipes and time-saved KPIs. Agents perform better when they call a real refactoring engine rather than rewriting text. JetBrains reports that Rider 2026.2.1's refactoring-code skill, which lets agents invoke ReSharper's syntax-tree operations, cut task time by 83% (157.9s to 26.6s) compared with text-based approximation. The same comparison showed 64% lower cost ($0.52 to $0.19) and 63% fewer tool calls. Apple's Xcode 27 ships framework-specific agentic refactoring skills written by framework maintainers, such as uikit-app-modernization and test-modernizer. Small, reviewable changes are the other pattern that works. Bloomberg's Pomona tool opens roughly 10-line PRs with a human in the loop. Of its PRs, 15 of 17 were merged, time-to-close was under two hours, and 80% of surveyed engineers were interested in adopting it. CodeScene's Deterministic PR Refactoring Agent reached general availability. It steers changes by measurable Code Health signals rather than by probabilistic prompting. Code health is emerging as the gate for scaling agents. CodeScene argues for a deterministic CodeHealth check on a 1.0–10.0 scale and treats code above 9.5 as AI-ready. Its study of 5,000 real programs refactored by six LLMs found that AI-generated changes fail at least 60% more often in unhealthy code. Only code scoring 7.0 or above was tested. CodeScene puts average hotspot CodeHealth in the Industrial & Technology sector at 5.15. Governed adoption can hold quality while agent use climbs. loveholidays kept 94% of teams above 9.75 CodeHealth while agent-assisted commits reached 80%. It also reports 20-30% year-on-year growth in deployment frequency after wiring CodeScene's CodeHealth MCP into agentic refactoring loops. Governance of this kind remains the exception rather than the norm. Code structure also shows up on the agent bill. Sonar's controlled study of 540 Claude Code runs found that cleaner codebases used 7.2% fewer input tokens and 8.5% fewer output tokens. Agents also revisited files 34% less often, and task completion was unchanged. That makes refactoring a measurable cost control for agentic workflows, not only a maintainability concern. Benchmarks set a hard ceiling on autonomous refactoring at repository scale. On Scale AI's SWE Atlas refactoring leaderboard, Claude Opus succeeds on 48.57% of tasks. SWE-Bench ProMax, with 170 multilingual instances, has GPT-5.2 at 41.2% on coordinated multi-file refactoring, far below the 75%+ scores seen on SWE-bench Verified. On SWE Refactor Bench's whole-repository migrations, only 5.4% of runs pass all three evaluation stages. The top performer, Claude Opus 5, scores 47/100. Code-smell repair and code deletion expose narrower weaknesses. SmellBench's 294-case evaluation found that the strongest configuration, Qwen Code with Claude Sonnet 4.5, scored only 50.34, and cross-file smells were the hardest. A study of deletion avoidance found that 29% of passing model patches wrapped logic in conditionals rather than removing it. Adding removal-only tests to SWE-bench tasks cut success from 63.2% to 41.9%. Green CI does not certify a refactoring. One developer refactored 100 Python functions with Claude Code, and all passed CI. Yet 7 regressed in production, with p95 latency drifting from 180ms to 240ms. Research on multi-turn programming found that 40-73% of tasks lose earlier correctness as requirements evolve. Re-running prior tests as a Verification Gate lifted quality from 75.8% to 87.9%. Practitioner guardrails such as scope locks and characterisation tests raised refactoring accuracy from 40% to 89%. Refactoring is disappearing from everyday development. GitClear's four-year analysis reports that refactoring fell from 21% of changed lines in 2022 to 3.8% in 2026, while copy-paste rose from 8.3% to 15.7%. GitClear also reports duplicate code blocks up 81% since AI went mainstream. WebProNews cites a study of 6,000+ public repositories that found 89.3% of AI-introduced issues were code smells. That study also found 22.7% of AI-introduced problems persisting in the latest version of each repository. Review scores are a poor predictor of downstream debt. New Relic's survey of 200 US technology decision-makers found that 94% rate AI code higher quality than human code at review. Yet 78% report more incidents, and 74% say at least 25% of AI code needs significant rework. In the same survey, 67% say AI generates or significantly refactors 51-75% of weekly code output. 62% say teams often ship it without line-by-line verification. Teams name verification capacity, not code generation, as the constraint. Qodo's survey of 500 developers and 300 engineering leaders found that 89% of organisations had experienced an AI-related production incident. Only 3.7% of leaders consider their existing processes sufficient, and 70% say code review now misses architectural context. A BairesDev survey of 705 developers found that 67% spend more time reviewing AI output. Production results trail benchmark claims. Telemetry from 12,400 agent runs shows 53% success in enterprise deployment, against 71% claimed in academic benchmarks. Compounding tool failures drag five-step pipelines down to 36.2% end-to-end success. A systems-level synthesis finds commits rising 180% while releases rise only 30%. It places the Verification Tax on review, testing and operations rather than on model capability. Technical debt is now eroding AI returns across whole portfolios. IBM's Institute for Business Value reports average AI ROI of 17% from a survey of 1,250 IT executives. Almost 70% of executives expect technical debt to make some AI initiatives financially untenable. IBM finds that organisations including modernisation and debt remediation in AI business cases project almost 30% higher AI ROI. Practicallogix reports that only 5-8% of enterprises achieve measurable at-scale ROI, and 73% of those winners restructured their processes. The blockers to broader adoption are organisational rather than technical. Visional's BizReach traces its structural debt to a mismatch between business understanding and implementation. It remediated that debt through an architecture committee, RFCs and a strangler-pattern migration, and finished separating its candidate search API. Its next step is a harness that lets AI attempt changes, receive verification results and retry in isolated environments. Humans keep decisions on business rules. Until verification capacity, codebase context and review discipline scale, governed refactoring remains a minority practice. ## Tier History - Research: 2023-06-01 – 2024-01-01 - Bleeding Edge: 2024-01-01 – present ## Evidence (172) - **2026-09-25** — [AI-Generated Code Grades Higher in Review, Yet Triggers Rise in Production Incidents](https://www.devopsdigest.com/ai-generated-code-grades-higher-review-yet-triggers-rise-production-incidents) (adoption-metric) New Relic survey of 200 US tech leaders: 94% rate AI code higher at review, yet 78% see more incidents and 74% say at least 25% needs significant rework. Evidence that 'agent debt' builds up after merge. - **2026-09-24** — [State of AI Code Quality Report](https://www.qodo.ai/state-of-ai-code-quality-report/) (adoption-metric) Qodo survey of 500 developers and 300 leaders: 89% had an AI-related production incident and only 3.7% of leaders think current processes are sufficient. Verification capacity is the named bottleneck. - **2026-09-21** — [AI Now Writes Half the Code. The Bill for Fixing It Is Just Arriving](https://www.webpronews.com/ai-now-writes-half-the-code-the-bill-for-fixing-it-is-just-arriving/) (news-coverage) Roundup adding new debt-persistence data: 89.3% of AI-introduced issues across 6,000+ repositories were code smells, and 22.7% persisted. Also a BairesDev survey of 705 developers on review load. - **2026-09-21** — [10 Lessons for Enterprise Code Modernization with OpenRewrite & Moderne](https://www.youtube.com/watch?v=c_c-JD5ZC-Y) (conference-talk) Practitioner talk from the vendor's channel, drawn from one to two years of enterprise OpenRewrite rollouts across hundreds of repositories. Covers platform teams, finder-recipe scoping, KPIs and where AI agents help. - **2026-09-18** — [技術的負債を長期化させない、AI時代の投資](https://engineering.visional.inc/blog/808/technical-debt-ai-investment/) (conference-talk) Visional/BizReach separated its candidate search API using the strangler pattern. It plans a verification-driven AI harness in which humans keep decisions on business rules. - **2026-09-15** — [IBM says cloud costs and tech debt erode AI returns](https://techinformed.com/ibm-says-cloud-costs-and-tech-debt-erode-ai-returns/) (adoption-metric) IBM IBV: almost 70% of executives expect technical debt to make some AI initiatives untenable. Including debt remediation in AI business cases projects almost 30% higher ROI. - **2026-09-15** — [Why AI Coding Agents Need a Deterministic Code Health Gate](https://codescene.com/blog/deterministic-code-health-gate-for-ai-agents) (opinion) CodeScene (vendor) study of 5,000 programs refactored by six LLMs: AI changes fail at least 60% more often in unhealthy code. Argues for a deterministic CodeHealth gate before scaling agents. - **2026-09-13** — [AI Agent & FinOps Statistics 2026: Production Failure Rates, Token Costs, and Enterprise Benchmarks](https://agenticspulse.com/posts/ai-agent-finops-statistics-2026.html) (adoption-metric) Production telemetry from 12,400 agentic engineering agent runs (Jan–Sept 2026) reveals 18.2pp reliability gap between academic benchmarks (71%) and real-world enterprise deployments (53%); tool failure 18.4%, compounding failures reduce 5-step pipeline success to 36.2%. Establishes fundamental research-to-practice gap in agent capability assessment. - **2026-09-09** — [Modernizing complex legacy code with AI agents - Mistral](https://mistral.ai/news/legacy-code-modernization/) (case-study) Named European energy operator successfully migrated 40,000 lines of physics-intensive Fortran 77 legacy code (reservoir simulator with no test suite) to C++ via AI agents with human-in-the-loop verification for numerical parity. Demonstrates bounded architectural refactoring at scale with governance discipline preventing autonomous drift. - **2026-09-04** — [Beyond Code Generation: Reliability, Verification, and Cost Economics in the Agentic Software Development Lifecycle](https://commonplace.workforcefutures.net/paper/arxiv:2609.04681) (research-paper) Systems-level synthesis of peer-reviewed research and production telemetry (2024–Sept 2026) quantifying Verification Tax as core bottleneck in agentic SDLC; shows commits +180%, projects +50%, but releases only +30%, establishing that verification/review capacity limits real-world deployment value. - **2026-09-04** — [Using Antigravity CLI to streamline dual-write database migration](https://cloud.google.com/blog/topics/developers-practitioners/using-antigravity-cli-to-streamline-dual-write-database-migration) (case-study) Google Finance Engineering case study: dual-write database migration from legacy datastore to Cloud Spanner in production financial system. Scale: 30+ DAOs with identical code changes via automated refactoring pipeline using Antigravity CLI in headless mode. Outcomes: significant manual effort reduction, high data migration fidelity, deterministic pattern adherence. - **2026-09-04** — [Refactoring Prompt Templates: What Actually Works](https://saaswithalex.pages.dev/posts/working-refactoring-prompt-templates/) (opinion) Practitioner empirical baseline on refactoring reliability: out-of-the-box LLMs achieve 40% accuracy on complex refactoring tasks; with guardrails (scope lock, tests required, diff-before-edit) accuracy rises to 89%. Documents three proven patterns for production-safe refactoring: behavior preservation discipline, characterization tests, incremental verification. - **2026-09-02** — [AI ROI 2026: Why 5-8% Ship + How to Join Them](https://www.practicallogix.com/the-2026-ai-roi-reckoning-why-5-8-ship-and-how-to-join-them/) (adoption-metric) Cross-analyst synthesis (BCG, KPMG, MIT NANDA, PwC) finding only 5-8% of enterprises achieve measurable at-scale AI ROI despite 44% enterprise-wide adoption. Critical insight: workflow redesign distinguishes success (73% of high performers redesigned workflows) from failure (95% report zero P&L impact). Governance adoption remains the tier-limiting factor. - **2026-09-01** — [Coding agents fail 94.6% of whole-repo migrations in new benchmark](https://agentry.news/coding-agents-fail-946-of-whole-repo-migrations-in-new-benchmark) (news-coverage) SWE Refactor Bench empirical evaluation on 20 full-repository migrations across 8 frontier models (520 runs): only 5.4% pass all validation stages (Migration Audit, Behavioral Tests, Agentic Verification). Critical negative signal—autonomous whole-repository refactoring remains beyond frontier capability despite individual-file success. - **2026-08-28** — [How Enterprises Use AI Code Refactoring to Reduce Technical Debt](https://wizr.ai/blog/ai-code-refactoring-for-enterprises/) (opinion) Wizr enterprise refactoring guide synthesizes industry metrics—McKinsey estimates technical debt equals 20-40% of technology estate value; Stripe finds developers lose 42% of weekly time to debt; Gartner reports 75% of engineers say AI generates new debt. Identifies five-step refactoring governance loop (Analysis, Detection, Suggestion, Validation, Review) as foundational for sustained adoption. - **2026-08-24** — [SWE Refactor Bench: Can Coding Agents Complete a Long-Horizon, Whole-Repository Stack Migration?](https://www.alphaxiv.org/abs/2608.23564) (research-paper) SWE Refactor Bench rigorously evaluates agentic refactoring on 20 full-repository stack migrations across 8 frontier models (520 runs). Only 5.4% pass all three evaluation stages (Migration Audit, Behavioral Tests, Agentic Verification); Claude Opus 5 achieves 47/100 score, exposing fundamental gap between test-pass metrics and production refactoring reliability. - **2026-08-23** — [AI Coding Agents Fail Large-Scale Refactoring: SWE-Bench ProMax Insights](https://appsecuritystandards.org/blog/ai-coding-agents-can-t-refactor-at-scale) (research-paper) SWE-Bench ProMax (170 multilingual instances across 7 languages) shows frontier models achieve only 41.2% success vs. 50-60% on earlier benchmarks, revealing significant capability ceiling. Root causes—dependency traversal failures, type system reasoning gaps, test-driven validation gaps—establish hard limits on production-scale refactoring autonomy. - **2026-08-20** — [Automating Code Modernization with OpenRewrite - Adyen](https://www.adyen.com/knowledge-hub/how-we-automated-code-modernization-with-openrewrite) (case-study) Adyen deployed OpenRewrite at enterprise scale across hundreds of developers and thousands of modules; first 2 months produced 4,000 automated MRs with 70% merge rate and <2hr median review turnaround. Critical orchestration insight—background cleanup before agent application reduced review fatigue, demonstrating deterministic refactoring automation viability. - **2026-08-19** — [Rider Hands AI Agents The Keys To Its Refactoring Engine For Safer, Faster, And Cheaper Results](https://blog.jetbrains.com/dotnet/2026/08/19/rider-refactoring-code-skill/) (product-ga) JetBrains Rider 2026.2.1 GA introduces refactoring-code skill enabling agents to invoke IDE refactoring operations directly via ReSharper syntax tree. Evaluation across 15 C# tasks shows 83% time reduction (157.9s→26.6s), 64% cost reduction ($0.52→$0.19), 63% fewer tool calls. Demonstrates semantic refactoring outperforms text-based approximation. - **2026-08-19** — [Code quality and AI agent bills: what practitioners need to know](https://nhimg.org/community/ai-beyond-identity/code-quality-and-ai-agent-bills-what-practitioners-need-to-know/) (adoption-metric) Sonar's controlled study of 540 Claude Code runs across matched repository pairs shows cleaner codebases use 7.2% fewer input tokens, 8.5% fewer output tokens, revisit files 34% less frequently. Task completion rates unchanged—benefit is purely operational efficiency, establishing code structure as measurable AI infrastructure cost control. - **2026-08-18** — [Code Copying Outpaces Refactoring in AI-Driven Development](https://www.linkedin.com/posts/albertsantalo_developers-are-now-five-times-more-likely-activity-7495475965037506560-V2zD) (adoption-metric) GitClear's 4-year longitudinal analysis of 623M code changes (2023–2026) documents systemic cultural shift; refactoring collapsed from 21% (2022) to 3.8% (2026) while copy-paste rose 81%, code duplication up 81%, error-masking up 47%. Developers now 5× more likely to copy-paste than refactor, inverting pre-AI development discipline. - **2026-08-14** — [AI agent refactored 189 files in a 717k-line codebase for $2,430](https://bizstack.tech/ai-agent-refactored-189-files-in-a-717k-line-codebase-for-2430/) (case-study) Fully autonomous refactoring of 717k-line production TypeScript codebase; specification-first protocol with 14 refinement + 17 verification audit cycles detected 201 defects pre-deployment. Deployed with zero production bugs across 30+ sessions; cost $2,430, 3 days, 189 files changed. - **2026-08-12** — [Anthropic's 80% Prompt Cut Shows AI Creating Its Own Tech Debt](https://futurumgroup.com/insights/anthropic-80-prompt-cut-shows-ai-creating-its-own-technical-debt/) (news-coverage) Platform migration (Anthropic's July 24 system prompt 80% reduction) forces downstream refactoring on customers; new class of infrastructure-level technical debt. Customers absorb prompt restructuring, CLAUDE.md refactoring, eval suite rebuilding. Signals AI tooling itself creates recurring refactoring burden. - **2026-08-10** — [SWE-Bench ProMax: Benchmarking Agents on Large-Scale Multilingual Code Refactoring](https://www.alphaxiv.org/abs/2608.09802) (research-paper) SWE-Bench ProMax (170 rigorously curated refactoring instances across 7 languages) reveals frontier capability gap; GPT-5.2 achieves 41.2% success on production-scale refactoring despite 75%+ performance on earlier benchmarks. Demonstrates architectural refactoring remains severely limited. - **2026-08-10** — [loveholidays scales AI coding while increasing code quality](https://codescene.com/customers/loveholidays) (case-study) Named UK travel agent (loveholidays) scaled agent-assisted commits to 80% while maintaining elite code quality (94% of teams >9.75 CodeHealth). Governance model using CodeScene CodeHealth MCP in agentic loop prevented quality degradation. 20-30% YoY deployment frequency growth with stable change failure rate. - **2026-08-06** — [AI Technical Debt Grows Because Agents Won't Delete Code](https://www.generativelabs.com/insights/ai-agents-wont-delete-code) (research-paper) Peer-reviewed research (arXiv:2607.28887) quantifying AI deletion avoidance as core refactoring failure mode; 29% of models wrap logic in conditionals rather than removing code (Guard-and-Go hedge). On retrofitted removal tests, SWE-bench success collapsed 63.2% to 41.9%, explaining technical debt acceleration. - **2026-08-05** — [Moderne Named a Leader in the Inaugural Gartner Magic Quadrant for AI-Augmented Code Modernization Tools](https://www.fidelity.com/news/article/technology/202608051231BIZWIRE_USPR_____20260805_BW534929) (product-ga) Gartner inaugural Magic Quadrant naming Moderne as leader in AI-Augmented Code Modernization Tools; recognition of 10,000+ deterministic OpenRewrite recipes enabling large-scale audited refactoring. Signals ecosystem maturity and enterprise-readiness for production deployment. - **2026-08-05** — [Vibe Coding in Production: The Bill Arrives at 90 Days](https://xpansionit.com/blogs/vibe-coding-technical-debt-australia) (case-study) XpansionIT multi-case analysis of unreviewed AI-generated code outcomes; 8,000+ startups required rescue engineering by mid-2026. Technical debt 30-41% increase post-adoption, 90-day inflection point losing 20-30% sprint capacity. 65% of vibe-coded production apps had security issues. - **2026-07-30** — [Refactoring Just Got a Price Tag: Why Code Quality Is Now a Line Item on Your AI Bill](https://reptile.haus/journal/refactoring-roi-token-cost-code-quality-ai-2026/) (case-study) Controlled experiment (Thoughtworks CTO Giles Edwards-Alexander) quantifying refactoring ROI in AI-era: 83% input token reduction (159,564 → 27,360) by refactoring oversized Rust file (17,155 → 3,695 lines across 19 files). First strong evidence that code structure now has measurable monetary cost under AI workflows. - **2026-07-29** — [Developers are attached to tools because tools encode trust](https://stackoverflow.blog/2026/07/29/developers-are-attached-to-tools-because-tools-encode-trust/) (opinion) Stack Overflow editorial on why AI coding tools are breaking developer trust and SDLC processes; usage rose to 84% but trust fell to 29%, exposing code review and validation bottlenecks as the new limiting factor for AI-generated code adoption. - **2026-07-29** — [Code Faster Today … Fail Faster Tomorrow?](https://www.computer.org/publications/tech-news/trends/protect-code-from-broken-ai-software) (industry-report) IEEE Computer Society analysis of 304,362 verified AI-authored commits found >15% introduced quality, security, or maintainability issues; categorizes failures into workflow (silent defects), security (hallucinations), technical (logic errors), and human-computer interaction failures. - **2026-07-27** — [AI Code Analysis Benchmark Reports for Engineering Leaders](https://blog.exceeds.ai/ai-code-analysis-benchmark-reports/) (industry-report) Comprehensive synthesis of 2026 industry benchmarks (Cortex, LinearB, CodeRabbit) quantifying productivity-vs-quality paradox: PR volume +20% while incidents +23.5%; refactoring activity drops 39.9%; AI-generated code contains 322% more privilege escalation paths than human code. - **2026-07-22** — [How AI is Transforming Legacy Modernization](https://www.fullstack.com/labs/resources/blog/how-ai-is-transforming-legacy-modernization) (case-study) Multiple named deployments of AI-assisted code refactoring with specific ROI metrics: FinTech 40% reduction in estimated hours, top 15 insurer 50% efficiency gains, healthcare provider $12M direct savings and 85% defect reduction via AI-modernized legacy codebase migration. - **2026-07-22** — [Refactoring AI-Generated Codebases: A Step-By-Step Architecture Rescue Plan](https://ehga.org/refactoring-ai-generated-codebases-a-step-by-step-architecture-rescue-plan) (opinion) Systematic 4-phase framework for rescuing AI-generated codebases from architectural debt; cites 1-in-5 AI functions incorrect, 70% refactoring collapse, 81% duplication increase, 74% drop in legacy maintenance as evidence of scale and severity. - **2026-07-21** — [Checksum releases the State of AI Code 2026, revealing growing confidence in AI-generated code despite persistent production failures](https://finance.yahoo.com/technology/ai/articles/checksum-releases-state-ai-code-130000127.html) (adoption-metric) Quantifies the trust-vs-reality gap in AI-generated code adoption: 78% trust AI more than last year, yet 61% shipped production incidents in 90 days; 74% rolled back AI code; 64.8% say AI code requires more review time than human code. - **2026-07-21** — [Agentic AI for Legacy Modernization: Faster and Fundable](https://www.simform.com/blog/agentic-ai-for-modernization/) (case-study) Production deployments of agentic AI for code refactoring with named customers: Visma's 3-million-line .NET modernization achieved 40% effort reduction; NTT DATA achieved 50% time-to-market reduction; demonstrates agentic orchestration at production scale. - **2026-07-21** — [GitLab 19.2 Puts AI Agents to Work on the Security Backlog](https://www.infoq.com/news/2026/07/gitlab-19-2-ai-agents/) (product-ga) Major vendor (GitLab) releasing GA features for agentic automation of dependency remediation and security review — direct evidence of industry tooling maturing to address technical debt backlogs created by AI-assisted coding workflows. - **2026-07-14** — [Trust but Verify? Uncovering the Security Debt of Autonomous Coding Agents](https://arxiv.org/abs/2607.12428) (research-paper) Analysis of 16,112 file changes across 4,022 agent-generated PRs; 38.9% contain security code smells; 82.3% of detected smells are supply chain integrity issues; 81.1% of hardcoded credentials escaped both automated and human review prior to merge. - **2026-07-14** — [AI's Biggest Failure Is Hiding in Your Existing Codebase](https://youmind.com/landing/x-viral-articles/ai-coding-failure-legacy-codebase) (case-study) Real deployment: 330 merged PRs in 6 months (~90% AI-generated) on legacy logistics platform; success required three steps—document before prompt, define risk zones (80% boilerplate, 20% logic, 0% critical), measure before scale; demonstrates brownfield refactoring governance. - **2026-07-13** — [The Debt the Cleanup Crew Created](https://mrdecentralize.substack.com/p/the-debt-the-cleanup-crew-created) (opinion) Analysis of 6,275 mature-codebase repositories tracking AI-introduced issues; agentic code produces 1.7× more issues per PR; 110,000+ unresolved issues by February 2026, with accumulation outpacing remediation—indicating structural debt acceleration. - **2026-07-10** — [Do These Violent Delights Have Violent Ends? Measuring the Post-Merge Fate of Agentic Code](https://arxiv.org/html/2607.09902v1) (research-paper) Longitudinal study of 182 repositories tracking agentic code post-merge for one year (May 2025–May 2026); agentic code requires 46% higher corrective maintenance and 45% higher bug-fixing rate, with accumulation patterns driven by no-review merge decisions. - **2026-07-10** — [The AI Productivity Paradox in Software Engineering: A Systematic Review of Code Quality, Technical Debt, and Organizational Throughout (2024–2026)](https://prereview.org/en-us/preprints/doi-10.5281-zenodo.21288635) (research-paper) Systematic review of 34 empirical studies (2,847 developers); 58% report code quality degradation and 22.7% technical debt persistence; short-term gains (+27% weeks 1–4) collapse by week 8; mandatory quality gates prevent 67% of debt insertion. - **2026-07-10** — [How Datadog Used Claude and Cursor for Test-Driven Production Migration](https://www.infoq.com/news/2026/07/datadog-ai-production-migration/) (case-study) Datadog refactored production Stream Router from KV to PostgreSQL using Claude and Cursor; test-driven approach ensured correctness; modularity and parallel infrastructure enabled safe rollout; demonstrates agentic refactoring success with measured governance. - **2026-07-09** — [The Tech Debt AI Leaves Behind](https://www.linkedin.com/pulse/tech-debt-ai-leaves-behind-hiren-dhaduk-gw9gf) (opinion) Analysis of 302,000 AI-authored commits; 15%+ introduce issues, 22.7% persist in latest codebase versions; introduces 'intent debt' concept—reasoning existing only in prompts cannot be reconstructed for maintenance; 58.8% of prompt-model combinations lose accuracy on model upgrade. - **2026-07-02** — [Regression Accumulation in Multi-Turn LLM Programming Conversations](https://arxiv.org/html/2607.01855v1) (research-paper) Empirical multi-model study (26,016 turns, 6 models across 542 tasks) documenting 40-73% regression accumulation as requirements evolve; Verification Gate strategy mitigates by retesting new code against prior tests, improving from 75.8% to 87.9%. - **2026-06-30** — [AI Legacy Modernization: The Complete 2026 Guide](https://getunblocked.com/blog/legacy-code-modernization/) (case-study) Meta Engineering case study: agentic refactoring on data pipelines blocked by context debt (only 5% of modules had agent-usable context); raising coverage to 100% via decision records reduced tool calls 40%, validating institutional memory as refactoring bottleneck. - **2026-06-23** — [AI Code Quality Signal Graphs - GitClear Research](https://www.gitclear.com/industry_stats/ai_code_quality_signal_graphs) (adoption-metric) 4-year longitudinal study (623M code changes): critical negative signal—refactoring moved-lines DOWN 74%, code duplication UP 81%, legacy maintenance DOWN 74%, error-masking UP 47%—quantifying AI-era write-only mode and structural debt accumulation. - **2026-06-18** — [Agentic AI Is Reshaping the SDLC: AWS Summit 2026](https://www.efficientlyconnected.com/agentic-ai-sdlc-aws-summit-2026/) (case-study) Named organizations (Net Smart, Avis, Virgin Australia) deployed agentic AI for legacy modernization reporting 80% dev time reduction, 64% code review time reduction; single engineer completed year-long Angular migration in 3 months; 'spec-driven development' for repeatable multi-stack migrations. - **2026-06-11** — [Apple Goes Agentic: AI Week of June 4-11, 2026](https://dev.to/alexmercedcoder/apple-goes-agentic-ai-week-of-june-4-11-2026-13mm) (news-coverage) Apple Xcode 27 (beta June 8) ships agentic refactoring tools including framework-specific skills authored by maintainers (uikit-app-modernization, test-modernizer); local inference + opt-in cloud, platform-native infrastructure for refactoring automation. - **2026-06-10** — [AI agentic refactoring in production code: what breaks first?](https://nhimg.org/community/agentic-ai-and-nhis/ai-agentic-refactoring-in-production-code-what-breaks-first/) (case-study) Named case study: 1Password agentic refactoring on multi-million-line Go monolith achieving 20-30% improvement with governance constraints; identifies core failure modes (sequencing, invariants, context) solvable via executable manifests, not model capability. - **2026-06-05** — [From Custom Logic to APIs - Understanding and Recommending API Replacement Refactorings](https://arxiv.org/abs/2606.06912) (research-paper) Empirical study mining 166K commits across Java projects; proposes AKIRA framework achieving 90% recall/88% precision on API replacement refactoring detection, advancing state-of-art from 21% to 81% recall on external dataset. - **2026-06-04** — [SmellBench: Towards Fine-Grained Evaluation of Code Agents on Refactoring Tasks](https://arxiv.org/html/2606.05574v1) (research-paper) Benchmark (294 cases, 7 smell types) evaluating code agents on refactoring; best (Qwen Code + Claude Sonnet 4.5) achieved only 50.34 score, revealing agents struggle with cross-file smell detection and architectural understanding. - **2026-06-03** — [Pomona: Continuous Code Quality Improvement via Small, Automated Changes at Bloomberg](https://arxiv.org/html/2606.06752v1) (case-study) Bloomberg production deployment of Pomona agentic refactoring tool generating small (~10-line) PRs targeting code quality; 15/17 merged with <2hr median time-to-close; 8/10 surveyed engineers expressed adoption interest. - **2026-05-28** — [I Refactored 100 Functions With Claude. CI Was Green. Production Got Slower in 7 Spots.](https://dev.to/kenimo49/i-refactored-100-functions-with-claude-ci-was-green-production-got-slower-in-7-spots-1d6) (case-study) Real-world deployment showing AI refactoring risks; 100 functions refactored by Claude Code, CI passed, but 7 had production performance regressions (p95 drift 180ms→240ms) revealing patterns missed by unit tests. - **2026-05-28** — [Announcement - Deterministic PR Refactoring Agents](https://codescene.com/blog/deterministic-pr-refactoring-agents) (product-ga) CodeScene production-ready PR Refactoring Agent using deterministic Code Health metrics (vs probabilistic prompting) to guide agentic refactoring, limiting scope to PR-introduced degradations with human-in-the-loop review. - **2026-05-26** — [Are Agents Leaving Your Code Stupid - Predictive Refactoring for Post-Repair Hardening](https://openreview.net/forum?id=qScXRhVWRB) (research-paper) Peer-reviewed research (ACL ARR 2026) proposing Memory-Augmented Predictive Refactoring (MAPR) to harden agent-generated code post-repair; achieves 100% pass-to-pass with zero regressions across LLM backbones (MiniMax, Gemini, GPT-5-mini, Qwen). - **2026-05-24** — [What 12 Months of AI-Generated Pull Requests Taught My Engineering Team](https://dev.to/sonia_bobrik_1939cdddd79d/what-12-months-of-ai-generated-pull-requests-taught-my-engineering-team-3915) (case-study) Named platform team (12-month analysis of 4,200 PRs): 26-55% more code shipped but incident rate +31%; team solved via mandatory video walkthrough for AI changes and observability investment. - **2026-05-23** — [AI's tech debt is invisible — even to AI. I solved it at the architecture layer.](https://dev.to/amingin_ai/ais-tech-debt-is-invisible-even-to-ai-i-solved-it-at-the-architecture-layer-1nh1) (case-study) OSS maintainer documenting AI architectural debt: agent ignored 12 existing service patterns, solution was persistent project-memory graph pinned to every commit, guiding AI consistency. - **2026-05-23** — [The Secret AI Refactor Workflow Nobody Uses (But Should)](https://dev.to/hopkins_jesse_cdb68cfa22c/the-secret-ai-refactor-workflow-nobody-uses-but-should-1p6d) (opinion) Practitioner methodology: adversarial refactoring workflow using AI to identify code fragility rather than generate solutions; claimed 120 hours debugging time saved vs traditional review. - **2026-05-21** — [I Let AI Refactor My Legacy Code for 14 Days — The Data Surprised Me](https://dev.to/hopkins_jesse_cdb68cfa22c/i-let-ai-refactor-my-legacy-code-for-14-days-the-data-surprised-me-5000) (case-study) 14-day autonomous refactoring trial on Rust API gateway: agent succeeded on warnings and naming, failed on cross-module dependencies (context window collapse at 8 days, 15 open PRs). - **2026-05-20** — [Quality and Security Signals in AI-Generated Python Refactoring Pull Requests](https://arxiv.org/abs/2605.21453) (research-paper) Empirical study measuring quality and security impact of AI-generated Python refactoring PRs: 22.5% improve a quality attribute, but 24.17% introduce new violations; 73.5% merge rate despite trade-offs. - **2026-05-16** — [Understanding DORA's 'The ROI of AI-assisted Software Development'](https://zenn.dev/inspector/articles/dora-roi-of-ai-2026-yomitoki?locale=en) (industry-report) DORA ROI framework analysis: 35-40% productivity gain on greenfield, but only 10% on complex legacy refactoring (Stanford 100k dev study)—establishing bounded-task success threshold. - **2026-05-14** — [Mining Subscenario Refactoring Opportunities in Behaviour-Driven Software Test Suites: ML Classifiers and LLM-Judge Baselines](https://arxiv.org/abs/2605.14568v1) (research-paper) Peer-reviewed research on BDD refactoring detection: SBERT + XGBoost achieved F1=0.891 on 339-repo corpus (5.4M slices), identifying 692k recurring patterns and cross-org opportunities. - **2026-05-13** — [Novacomp Case Study - AWS](https://aws.amazon.com/solutions/case-studies/novacomp-case-study/) (case-study) Novacomp case study: Java 8→17 migration of 10k LOC completed in 50 minutes (vs 3-week estimate) with 60% technical debt reduction and zero regressions using Amazon Q Agent. - **2026-05-11** — [Developer Productivity Guide: Measurement and Metrics in 2026](https://gogloby.com/insights/developer-productivity-guide/) (adoption-metric) Quantifies technical debt accumulation: 38% see deployment frequency rise with change failure rate increase; 41% AI commits correlate with higher rework; PR review times spike 441% YoY. - **2026-05-09** — [SWE Atlas - Refactoring](https://labs.scale.com/leaderboard/sweatlas-refactoring) (research-paper) Frontier benchmark: Claude Opus achieves 48.57% success on 70 production refactoring tasks; open models lag with regressions, establishing hard capability limits for autonomous code restructuring. - **2026-05-07** — [SmellBench: Evaluating LLM Agents on Architectural Code Smell Repair](https://arxiv.org/abs/2605.07001v1) (research-paper) Empirical study: 63% false positive detection rate on architectural code smell repair; exposes autonomy-accuracy trade-off, confirming architectural refactoring beyond current capability. - **2026-05-06** — [Faster Code, More Failures. The AI Paradox](https://www.datapro.news/p/faster-code-more-failures-the-ai-paradox) (industry-report) METR randomized trial synthesis: developers perceive 20% speedup but measure 19% slowdown on complex systems; GitClear documents 8x duplication and 1.57x more security vulnerabilities. - **2026-05-04** — [AI-Generated Smells: An Analysis of Code and Architecture in LLM and Agent-Driven Development](https://arxiv.org/abs/2605.02741) (research-paper) Peer-reviewed research: 'Volume-Quality Inverse Law' proves code volume predicts structural degradation; AI produces 'machine signature of defects' invisible to functional testing. - **2026-04-30** — [.NET 8 Modernization: Legacy to Cloud Case Studies](https://www.netsolutions.com/insights/dotnet-8-modernization-case-studies/) (case-study) Named vendor case study: .NET Framework→.NET 8 modernization achieved 35% timeline reduction, 60% infrastructure cost savings, 50% API response time improvement. - **2026-04-30** — [89% of Enterprise Engineering Teams Have Experienced an AI-Generated Code Incident. The Data Explains Why.](https://www.qodo.ai/blog/ai-coding-paradox-report/) (adoption-metric) Censuswide survey (N=500): 89% experienced AI incidents; 25% suffered complete outages; 41% report increased manual review time post-AI adoption, documenting verification bottleneck. - **2026-04-28** — [How Blue Pearl modernized an outdated codebase and resolved a risky security posture with IBM Bob](https://www.ibm.com/new/product-blog/how-blue-pearl-modernized-an-outdated-codebase-and-a-resolved-a-risky-security-posture-with-ibm-bob) (case-study) Named deployment: Blue Pearl's Java 11→21 refactoring achieved 90% timeline compression (3 days vs 30+), 92% test coverage, 127 deprecated APIs resolved, zero CVEs post-migration. - **2026-04-25** — [OpenRewrite by Moderne | Large Scale Automated Refactoring](https://docs.openrewrite.org) (significant-repo) OpenRewrite documentation describing LST-based recipe engine for large-scale code migrations, framework updates, and consistency fixes—foundational tooling for bounded refactoring automation at scale. - **2026-04-24** — [SonarQube Review 2026: An Honest Take for Enterprise Engineering Teams](https://techsy.io/en/blog/sonarqube-review) (industry-report) Technical review of SonarQube 2026 covering AI CodeFix GA, AI Code Assurance, Quality Gates, and code quality framework—addresses maintaining code quality amid rapid AI-assisted development. - **2026-04-22** — [When Code Gets Cheaper, Judgment Gets More Precious: Quality Bottlenecks in Enterprise AI Systems](https://deepsense.ai/blog/when-code-gets-cheaper-judgment-gets-more-precious-quality-bottlenecks-in-enterprise-ai-systems/) (case-study) Enterprise case study of throughput trap: teams accumulate surface area faster than they can validate it; AI builds on dead code and ignores legacy patterns. Signal on organizational governance failure modes. - **2026-04-20** — [Vibe Code at Scale: Managing Technical Debt When AI Writes Most of Your Codebase](https://tianpan.co/blog/2026-04-20-vibe-code-at-scale-technical-debt) (case-study) Named incident (6.3M orders lost), 153M-line analysis showing 1.7x defect rate, 40% of AI code rewritten within 2 weeks, refactoring dropped 60%, 30-41% technical debt increase—identifies structural failure modes and management approaches. - **2026-04-18** — [Your Code Review Was Built for Humans. 41% of Code Isn't](https://www.iqsource.ai/en/blog/ai-code-review-quality-governance/) (industry-report) Structural failure of traditional CI/CD/review infrastructure under 41% AI-generated code adoption; prescribes new quality gates (behavioral testing, mutation testing, architectural consistency) for debt management governance. - **2026-04-15** — [AI-Generated Code in Production: What the Data Says in 2026](https://www.buildorbit.studio/blog/ai-generated-code-production-debugging-problems) (case-study) Production incident evidence: Amazon March 2026 incidents ($6.3M loss), Lightrun SRE survey (43% require manual debugging post-QA, 0% confident in deployment), CodeRabbit metrics (1.75x correctness errors). Critical negative signal. - **2026-04-15** — [CodeScene | Technology Radar | Thoughtworks United Kingdom](https://www.thoughtworks.com/en-gb/radar/tools/codescene) (industry-report) Thoughtworks Radar recognizes CodeScene as emerging tool for behavioral code analysis; explicitly valuable as guardrail for AI coding agent adoption to prevent technical debt introduction. - **2026-04-14** — [The AI Code Review Trap: Why Faster Reviews Are Making Your Codebase Worse](https://tianpan.co/blog/2026-04-14-ai-code-review-automation-bias) (industry-report) Code review process failures under AI adoption: automation bias allows 23.5% incident increase despite faster reviews, architectural debt accumulation at 322% baseline rate, skill erosion among junior engineers. Governance-critical signal. - **2026-04-12** — [The AI Engineering Report 2026: The AI Acceleration Whiplash](https://www.faros.ai/blog/ai-acceleration-whiplash-takeaways) (adoption-metric) Large-scale telemetry (22K developers, 4K+ teams) shows AI as primary code author with severe quality degradation: code churn +861%, bugs +54%, incidents +242.7%, confirming the problem refactoring practices must address. - **2026-04-09** — [KPMG Global Tech Report 2026: Bridging the AI Ambition-Execution Gap](https://www.libertify.com/interactive-library/kpmg-global-tech-report-2026-ai-ambition-execution-gap/) (industry-report) Analyst survey of 2,500 executives shows high-performer orgs achieve 4.5x ROI via disciplined tech debt management vs 2x average, identifying tech debt governance as key competitive differentiator. - **2026-04-07** — [AI vs. Debt: Stop Your Code from Becoming a Time Bomb](https://www.baytechconsulting.com/blog/ai-vs-debt-stop-code-time-bomb) (industry-report) Meta-analysis using DORA, GitClear (211M LOC), METR research quantifies productivity paradox: code churn doubled (3.1%→5.7%), code cloning 4x (8.3%→12.3%), refactoring collapsed (25%→<10%) despite velocity claims. - **2026-04-06** — [Copilot AI Code Insertion Security Risks: Team Governance Playbook](https://branch8.com/posts/copilot-ai-code-insertion-security-risks-team-governance) (case-study) Real-world incident: Copilot-generated authentication code exposed tokens; code review missed vulnerability due to AI-assisted reviewers. Documents governance barriers and Stanford research on developer confidence gap. - **2026-04-06** — [88% of Developers Report Negative AI Impact on Technical Debt—Yet We Ship It Anyway](https://tianpan.co/forum/t/88-of-developers-report-negative-ai-impact-on-technical-debt-yet-we-ship-it-anyway-are-we-trading-q1-velocity-for-q3-crisis/4283) (adoption-metric) Survey reveals 88% report negative AI impact on debt; 30-41% debt increase within 90 days. Identifies sustainable threshold (25-40% AI code) and documents governance framework with tiered review and 20% sprint debt budget. - **2026-04-05** — [Technical Debt Increased 30-41% After AI Adoption. By Year Two, Our Maintenance Costs Hit 3.8x](https://tianpan.co/forum/t/technical-debt-increased-30-41-after-ai-adoption-by-year-two-our-maintenance-costs-hit-3-8x-when-does-ship-faster-become-pay-forever/4149) (case-study) Detailed deployment metrics reveal Year 1 velocity gains (40% faster) reversed in Year 2 by 3.8x maintenance costs and 60% reduced refactoring. Includes governance framework (35% AI cap, 20% sprint debt budget) that resolved crisis. - **2026-04-04** — [Engineering AI Adoption: 2026 Framework to Measure ROI](https://blog.exceeds.ai/engineering-effectiveness-ai-adoption-2026/) (adoption-metric) Quantifies productivity paradox: 91% team adoption but only 15% achieve business value; METR research shows 19-20% longer task times in some contexts, revealing hidden technical debt costs masked by velocity metrics. - **2026-04-03** — [AI-Assisted Unit Test Writing and Test-Driven Code Refactoring: A Case Study](https://papers.cool/arxiv/2604.03135) (research-paper) Peer-reviewed case study demonstrates LLM-assisted test generation (16K test lines in hours vs weeks) enabling safe refactoring with 78% branch coverage and reduced regression risk through test-driven constraints. - **2026-04-02** — [CodeScene Review: Behavioral Code Analysis That Links Code Health to Business Impact](https://aicoolies.com/reviews/codescene-review) (industry-report) Independent review validates CodeScene 6x more accurate than SonarQube for maintainability prediction; new MCP integration enables AI agents to check code health in real-time, preventing AI-generated debt. - **2026-04-02** — [AI Governance Roadmap for Engineering Leaders in 2026](https://blog.exceeds.ai/ai-governance-roadmap-engineering-2026/) (adoption-metric) Quantifies AI-specific technical debt: 30-50% annual maintenance vs 20-25% traditional code. Organizations with shadow AI suffered +$670K breach costs; 7-phase governance roadmap emphasizes code-level visibility and outcome tracking. - **2026-03-31** — [Fix Pull Request Issues with SonarQube Remediation Agent](https://www.sonarsource.com/resources/library/fix-pull-request-issues-with-the-sonarqube-remediation-agent/) (product-ga) Product GA for AI-powered automatic code fix generation with built-in sandbox verification and human review gates, deployed across Java, JavaScript, TypeScript, Python at scale. - **2026-03-28** — [AI Technical Debt: $2.4T Cost Enterprises Can't Ignore](https://byteiota.com/ai-technical-debt-2-4t-cost-enterprises-cant-ignore/) (adoption-metric) Macro-level analysis: technical debt costs US $2.4T annually; high-debt orgs spend 40% more on maintenance. Organizations managing debt in business cases project 29% higher ROI. - **2026-03-26** — [Engineering Team AI Coding Tools ROI Metrics Guide](https://blog.exceeds.ai/ai-coding-productivity-metrics-roi/) (adoption-metric) Quantifies technical debt burden: AI code shows 1.7× higher defect density, 23.5% more incidents, 39% higher complexity requiring refactoring; proposes tracking methodology for debt management. - **2026-03-24** — [Spring 2026 GenAI Code Security Update - Veracode](https://www.veracode.com/blog/spring-2026-genai-code-security/) (industry-report) Longitudinal testing of 150+ models reveals security debt: 55% pass rate (45% introduce vulnerabilities) despite syntax improving to 95%, showing persistent quality gap in AI-generated code requiring refactoring. - **2026-03-18** — [Making Legacy Code AI-Ready: Benchmarks on Agentic Refactoring](https://codescene.com/blog/making-legacy-code-ai-ready-benchmarks-on-agentic-refactoring) (case-study) Empirical deployment evidence: Claude Code with CodeHealth MCP achieved 2–5x improvement on 25k files; Extract Method refactorings increased 3x with structured guidance; industry baseline code health 5.15/10 vs AI-safe requirement 9.4+. - **2026-03-16** — [AI Technical Debt: 30-41% Increase Hits Developers](https://byteiota.com/ai-technical-debt-30-41-increase-hits-developers/) (adoption-metric) Independent analysis of 8.1M pull requests from 4,800 teams shows AI-generated code has 1.7× more issues, 30-41% technical debt increase, and 19% slower delivery, quantifying deployment impact. - **2026-03-11** — [Why AI is both the problem and the cure for legacy code](https://engineeringharmony.substack.com/p/why-ai-is-both-the-problem-and-the) (case-study) Deployment case studies show AI refactoring wins: Airbnb migrated 3.5k React tests in 6 weeks (1.5-year estimate, 75%→97% success); monday.com completed JS monolith breakup in 6 months (8-year estimate). - **2026-03-10** — [AI Technical Debt: How AI-Generated Code Creates Hidden Costs – Tembo](https://www.tembo.io/blog/ai-technical-debt) (industry-report) Framework distinguishing AI technical debt (invisible by default, scales with adoption). GitClear 211M LOC analysis: refactoring dropped 25%→<10% (2021-2024); code churn +5.5%→7.9%; copy-paste code 8.3%→12.3%. - **2026-03-04** — [CodeTaste: Can LLMs Generate Human-Level Code Refactorings?](https://arxiv.org/abs/2603.04177) (research-paper) Peer-reviewed benchmark study of LLM agents on 100 real-world refactoring tasks reveals critical gap: 70% accuracy with detailed instructions vs <8% autonomous refactoring success, establishing limits of unguided AI refactoring. - **2026-02-23** — [How AI Turned a 2-Year Project Into a 1-Week Sprint](https://www.holgerscode.com/blog/2026/02/23/adapt-or-disappear-how-ai-turned-a-2-year-project-into-a-1-week-sprint/) (case-study) Delphi 7 to TypeScript/Next.js migration completed in one week, generating 50,000 lines of code with AI pattern replication. Demonstrates bounded refactoring success combining human expertise and AI execution at scale. - **2026-02-23** — [I Let AI Rewrite 40% of My Codebase. Here's What Actually Happened](https://dev.to/techstratos/i-let-ai-rewrite-40-of-my-codebase-heres-what-actually-happened-1jd6) (case-study) Production refactoring of 60K lines over 3 months: 30-40% feature speed improvement, 18% test coverage increase, but significant failures (async errors, performance regressions, architectural drift). Mixed outcomes highlighting AI amplifies senior judgment. - **2026-02-22** — [GitHub Debt Insights Predicted Our Production Incident 3 Weeks Early](https://tianpan.co/forum/t/github-debt-insights-predicted-our-production-incident-3-weeks-early-heres-what-we-learned/1128) (case-study) AI tool predicted authentication module incident 3 weeks early. After 3 months: 45% reduction in tech debt incidents, 18% reduction in unplanned work, 35% reduction in bug fix time. Demonstrates AI-assisted detection effectiveness. - **2026-02-19** — [Moderne Adds Python Support to Tame AI-Driven Technical Debt](https://briefglance.com/articles/moderne-adds-python-support-to-tame-ai-driven-technical-debt) (product-ga) Moderne extends automated refactoring platform to Python, addressing escalating AI-generated technical debt with deterministic transformations. Industry metrics: 50% of code changes now AI-generated, developers spend one-third of time on debt. - **2026-02-18** — [Mind the gap: Closing the AI trust gap for developers](https://stackoverflow.blog/2026/02/18/closing-the-developer-ai-trust-gap/) (industry-report) Stack Overflow survey: 84% adoption but only 29% trust AI. Defines trust as willingness to deploy with minimal review; links to technical debt risk. Critical signal on adoption-trust divergence limiting widespread refactoring automation. - **2026-02-12** — [The great toil shift: How AI is redefining technical debt](https://www.sonarsource.com/blog/how-ai-is-redefining-technical-debt/) (industry-report) SonarSource survey: 88% of developers report negative AI impacts (53% unreliable code, 40% duplication), while 93% report positive impacts. Captures dual nature of AI on technical debt: productivity gains paired with new debt creation. - **2026-01-31** — [AI Tools, Agents, and the Future of Software Development](https://marionoioso.com/2026/01/31/genai-agents-legacy-modernization/) (industry-report) Consulting firm synthesis (Thoughtworks, BCG, EY, IBM, Xebia) documenting GenAI modernization approaches including BCG case processing 3M lines of COBOL in days, showing 30-60% efficiency gains in legacy system refactoring initiatives. - **2026-01-30** — [Software Development Security, Technical Debt & AI Governance](https://www.baytechconsulting.com/blog/security-technical-debt-ai-governance-2026) (opinion) Critical analysis framing rapid AI code generation as creating 'Efficiency Paradox' where initial speed masks high-interest technical debt accumulation in system integration, security hardening, and edge cases—amplifying debt for junior developers without expert guidance. - **2026-01-30** — [Day 30: Managing Technical Debt When Shipping Fast](https://31daysofvibecoding.com/2026/01/30/technical-debt/) (tutorial) Practitioner guide with specific debt management strategies for AI-assisted development: Debt Inventory Prompt, 20% Rule for allocation, AI-assisted debt paydown patterns, and prevention tactics to combat inconsistency and shallow testing in AI-generated code. - **2026-01-14** — [I Tried to Prove AI Was Sloppy. It Ended Up Refactoring My Career](https://blog.d3alpha.com/blog/2026-01-14-i-tried-to-prove-ai-was-sloppy-it-ended-up-refactoring-my-career) (case-study) Production case study: AI agent successfully refactored and rebuilt complex enterprise data flow system from schema and documentation in weeks versus prior 76-day manual project, demonstrating capability for large-scale refactoring with iterative validation. - **2026-01-01** — ["TODO: Fix the Mess Gemini Created": Towards Understanding GenAI-Induced Self-Admitted Technical Debt](https://conf.researchr.org/details/TechDebt-2026/TechDebt-2026-main/2/-TODO-Fix-the-Mess-Gemini-Created-Towards-Understanding-GenAI-Induced-Self-Admitte) (research-paper) Peer-reviewed TechDebt 2026 research analyzing 6,540 LLM-referencing code comments, identifying 81 cases of GenAI-Induced Self-admitted Technical Debt with developers expressing uncertainty about AI-generated code quality and delayed verification. - **2026-01-01** — [Scale OpenRewrite auto-refactoring with Moderne](http://www.moderne.ai) (product-ga) Moderne's AI-powered multi-repo auto-refactoring platform reaches enterprise customer adoption (MEDHOST, Interactions, Allstate, Intel Capital, Choice Hotels) with deterministic large-scale transformations across thousands of repositories. - **2025-12-29** — [Developers remain willing but reluctant to use AI: The 2025 Developer Survey results are here](https://stackoverflow.blog/2025/12/29/developers-remain-willing-but-reluctant-to-use-ai-the-2025-developer-survey-results-are-here/) (adoption-metric) Stack Overflow survey of 49,000+ developers: 80% use AI tools for coding, but trust fell to 29%; 45% cite 'almost right but not quite' AI solutions, 66% spend extra time fixing AI-generated code, confirming technical debt creation at scale. - **2025-12-01** — [How AI-Driven Refactoring Cut Legacy Code Migration to Just 4 Months](https://engineering.salesforce.com/how-ai-driven-refactoring-cut-a-2-year-legacy-code-migration-to-4-months/) (case-study) Salesforce production migration of Own Archive legacy codebase: AI-driven refactoring reduced 2-year manual effort to 4 months, modernizing 275 Apex classes and 3537 total files into Core infrastructure with fully native product delivery. - **2025-12-01** — [AI Technical Debt: The Silent Cybersecurity Crisis](https://www.forcepoint.com/blog/x-labs/ai-technical-debt) (industry-report) Forcepoint analysis linking rapid AI adoption to security-critical technical debt: Toyota and Decathlon breaches traced to legacy migration practices and misconfiguration, highlighting data risk consequences of unmanaged technical debt from AI-assisted deployments. - **2025-12-01** — [The AI Refactoring Trap: Why "Messy" Code is Often Better](https://dev.to/samuel_ochaba_eb9c875fa89/the-ai-refactoring-trap-why-messy-code-is-often-better-563p) (opinion) Developer critical assessment: AI over-applies DRY principles leading to wrong abstractions; React component example shows how AI-generated abstractions create 'Conditional Monster' refactoring debt, cautioning against unguided large-scale AI-driven refactoring. - **2025-10-23** — [AI-Assisted Test Automation: A Real-World Refactoring Case Study](https://altom.com/case-studies/ai-assisted-test-automation-a-real-world-refactoring-case-study/) (case-study) Altom consultancy pilot refactoring legacy video surveillance test suite with GitHub Copilot Agent (Claude Sonnet 4): AI accelerated debugging with precise guidance but struggled with multi-level inheritance, showing mixed deployment outcomes requiring human expertise. - **2025-10-21** — [The State of Developer Ecosystem 2025: Coding in the Age of AI](https://blog.jetbrains.com/research/2025/10/state-of-developer-ecosystem-2025/) (adoption-metric) JetBrains survey of 24,534 developers across 194 countries: 85% regularly use AI tools for coding, 62% rely on AI coding assistants; developers report near-universal time savings but 66% question productivity metric accuracy in technical debt contexts. - **2025-09-25** — [Gen AI tools trim java upgrade for tax software](https://www.cognizant.com/us/en/case-studies/gen-ai-tools-trim-java-upgrade-for-tax-software) (case-study) Cognizant case study of gen AI-led Java migration for market-leading tax software: 35% cost reduction, 25% effort reduction on major upgrade, demonstrating deployment success in bounded refactoring scenarios. - **2025-09-23** — [How are developers using AI? Inside our 2025 DORA report](https://blog.google/innovation-and-ai/technology/developers-tools/dora-report-2025/) (industry-report) Google DORA research surveying ~5,000 technology professionals on AI in software development practices, providing authoritative adoption metrics on how teams are integrating AI-assisted refactoring into development workflows. - **2025-08-27** — [Vibe Shift in AI Coding: Senior Developers Ship 2.5x More Than Junior Developers](https://www.fastly.com/blog/senior-developers-ship-more-ai-code) (adoption-metric) Fastly survey of 791 developers: 33% of senior developers (10+ years experience) ship >50% AI-generated code, 2.5x the rate of junior developers, indicating production deployment of AI refactoring at scale in experienced teams. - **2025-07-31** — [Engineer-Guided Legacy Code Refactoring with GenAI](https://www.ksolves.com/case-studies/ai-ml/legacy-telecom-modernization-genai) (case-study) European telecom leader modernization: phased, engineer-guided AI refactoring of monolithic legacy architecture with minimal documentation into modular services, showing success with constrained scope and oversight. - **2025-07-29** — [Developers remain willing but reluctant to use AI: The 2025 Developer Survey results are here](https://stackoverflow.blog/2025/07/29/developers-remain-willing-but-reluctant-to-use-ai-the-2025-developer-survey-results-are-here) (adoption-metric) Stack Overflow survey of 49,000+ developers: widespread AI adoption continuing despite growing distrust in output quality, capturing ecosystem sentiment on AI-generated code reliability for refactoring tasks. - **2025-07-17** — [What's Missing With AI-Generated Code? Refactoring](https://dev.to/_steve_fenton_/whats-missing-with-ai-generated-code-refactoring-20h6) (news-coverage) GitClear analysis of 211M lines of code: refactoring signals crashing while duplication and churn accelerating—2024 is first year code introduction exceeds refactoring activity, showing AI-generated code creates debt faster than it's remediated. - **2025-06-12** — [Despite 78% Claiming Productivity Gains, Two in Three Developers Say AI Misses Critical Context](https://www.prnewswire.com/news-releases/despite-78-claiming-productivity-gains-two-in-three-developers-say-ai-misses-critical-context-according-to-qodo-survey-302480084.html) (adoption-metric) Qodo survey of 600+ developers: 82% use AI assistants daily/weekly but two-thirds say AI misses critical context, indicating technical debt risks from contextual blindness in large refactoring tasks. - **2025-05-22** — [Refactoring with AI: Why "Fix Everything" Is a Terrible Prompt](https://adamtheautomator.com/coding-ai-break-down/) (tutorial) Developer tutorial documenting AI refactoring failures on real components (silent state-handling breakage across pages), providing practical guardrails and arguing for incremental, scoped refactoring over large-scale AI automation. - **2025-05-07** — [Which Platform Actually Connects Code Quality to Business Results?](https://www.codeant.ai/blogs/code-quality-platforms-business-results) (industry-report) CodeAnt analysis showing 40% deployment lead time improvement from reducing service complexity and documenting AI-code defect density (1.7x higher than human code), linking technical debt metrics to business velocity. - **2025-05-06** — [SonarQube for Federal Agencies: Complying with AI Policies in Code Development](https://www.sonarsource.com/resources/library/complying-with-ai-policies-in-code-development/) (industry-report) SonarQube AI Code Assurance workflow adopted by federal agencies to validate and auto-fix AI-generated code, demonstrating enterprise-grade tooling maturity for technical debt management in regulated environments. - **2025-05-06** — [Some Observations on AI/Agentic Refactoring](https://amanagrawal.blog/2025/05/06/some-observations-on-ai-agentic-refactoring/) (opinion) Practitioner experiment with production C# refactoring showing critical AI failures: type mismatches breaking code, unnecessary method renames, architectural misunderstandings—exposing limitations of autonomous refactoring. - **2025-04-23** — [What Web Developers Really Think About AI in 2025](https://dev.to/sachagreif/what-web-developers-really-think-about-ai-in-2025-2fjn) (adoption-metric) Survey of 4,000+ web developers: 91% use AI for code generation, but 61% of AI-produced code requires refactoring due to poor readability and excessive repetition, confirming technical debt creation at scale. - **2025-03-04** — [Why AI-generated code is creating a technical debt nightmare](https://www.okoone.com/spark/technology-innovation/why-ai-generated-code-is-creating-a-technical-debt-nightmare/) (news-coverage) Synthesis of GitClear/Harness/Google reports: AI code shows 8x code duplication rise, 10x redundancy since 2022, 7.2% delivery stability decrease, creating higher maintenance costs and defect rates. - **2025-02-11** — [D2DO255: Is AI the Magic Solution for Refactoring Legacy Code?](https://packetpushers.net/podcasts/day-two-devops/d2do255-is-ai-the-magic-solution-for-refactoring-legacy-code/) (conference-talk) GitLab Federal CTO describes AI as 'coach, not magic wand' for legacy refactoring; notes context gaps limit LLM effectiveness and traditional engineering techniques remain superior for reliability. - **2025-01-06** — [AI and the $2.41 Trillion Technical Debt](https://opentools.ai/news/ai-and-the-dollar241-trillion-technical-debt) (industry-report) Accenture research: US technical debt costs $2.41 trillion annually, blocking AI adoption; companies investing 15% of IT budget in remediation plus AI assistance achieve 60% higher revenue growth. - **2025-01-01** — [Comparing SonarQube Server 10.2 with 2025.1 LTA](https://www.sonarsource.com/products/sonarqube/feature-comparison/10-2/) (product-ga) SonarQube 2025.1 LTA releases AI Code Assurance and AI CodeFix features for validating and auto-fixing AI-generated code, addressing ecosystem shift toward managing AI-created technical debt. - **2025-01-01** — [Sonar's State of Code Developer Survey 2025](https://www.sonarsource.com/jp/resources/white-papers/) (industry-report) Survey of 1,100+ enterprise developers: 72% of AI users run it daily but 96% don't fully trust output, only 48% verify before commit. Gap between adoption and oversight creates mounting technical debt risks. - **2025-01-01** — [The State of AI Coding 2025](https://stateof.themodernsoftware.dev) (adoption-metric) Survey of 195 developers: 98% use AI tools multiple times weekly; nearly 40% report AI generating >50% of their codebase monthly, demonstrating pervasive reliance creating technical debt management urgency. - **2024-12-30** — [Technical debt in AI-enabled systems: On the prevalence, severity, and management...](https://oulu.cris.fi/en/publications/e319729b-996f-42d7-9e58-f5d6d4701a09) (research-paper) Peer-reviewed empirical study in Journal of Systems and Software surveying prevalence, severity, and management of technical debt in AI-enabled systems, providing quantitative adoption metrics. - **2024-12-20** — [Trust Calibration in IDEs: Paving the Way for Widespread Adoption of AI Refactoring](http://www.arxiv.org/abs/2412.15948) (research-paper) arXiv position paper on safeguards for AI refactoring in IDEs, addressing LLM risks (breaking changes, security vulnerabilities) and proposing trustworthy guardrails for widespread adoption. - **2024-11-26** — [Artificial Intelligence and Technical Debt: Navigating the New Frontier](https://www.castsoftware.com/pulse/artificial-intelligence-and-technical-debt-navigating-the-new-frontier) (industry-report) CAST analysis of AI-generated code's dual impact on technical debt: accelerates development but risks inconsistencies, maintenance challenges, and security vulnerabilities without proper oversight and intelligence tools. - **2024-11-14** — [AI makes tech debt more expensive](https://news.ycombinator.com/item?id=42137527) (opinion) Practitioner discussion: LLMs perform well on standard patterns but worsen technical debt in legacy/novel codebases; companies with young, high-quality code benefit most, while gnarly legacy systems face adoption barriers. - **2024-11-09** — [Can AI Truly Refactor Your Code Better Than You? - LLMs in Software Development](https://theministryofai.org/can-ai-truly-refactor-your-code-better-than-you-discover-the-pros-and-cons-of-llms-in-software-development) (news-coverage) Analysis of empirical refactoring study: ChatGPT 63.6% matching expert-quality refactorings, Gemini 56.2%; excels at inline/extraction but struggles with naming; safety concerns remain for production use. - **2024-11-07** — [Stack Overflow's 2025 Developer Survey - Trust in AI at an All Time Low](https://stackoverflow.co/company/press/archive/stack-overflow-2025-developer-survey) (adoption-metric) Survey of 49,000 developers: 84% use or plan to use AI tools (up from 76%), but trust declining (46% distrust output, up from 31%), and 45% find AI-generated code debugging time-consuming. - **2024-09-23** — [Where developers feel AI coding tools are working—and where they're missing the mark](https://stackoverflow.blog/2024/09/23/where-developers-feel-ai-coding-tools-are-working-and-where-they-re-missing-the-mark/) (adoption-metric) Stack Overflow survey of 65K+ developers: 76% adoption of AI coding tools (up from 70%), but only 42% trust output and 45% report tools inadequate for complex tasks. - **2024-09-17** — [Businesses thought AI could write code. now they're playing whack-a-mole with outages](https://techtonicshifts.blog/2024/09/17/businesses-thought-ai-could-write-code-now-theyre-playing-whack-a-mole-with-outages/) (opinion) Critical synthesis of AI code correctness research: ChatGPT 65.2%, Copilot 46.3%, Amazon CodeWhisperer 31.1%; code churn doubling by 2024; over half of companies report security issues from AI-generated code. - **2024-09-06** — [Exploring the Advances in Using Machine Learning to Identify Technical Debt and Self-Admitted Technical Debt](https://www.arxiv.org/abs/2409.04662) (research-paper) Peer-reviewed literature review (IEEE SERA 2025) finding BERT models significantly more effective than alternatives for technical debt identification, providing empirical signal on ML maturity. - **2024-08-30** — [The State Of Technical Debt In The US, 2024 - Forrester](https://www.forrester.com/report/the-state-of-technical-debt-in-the-us-2024/RES181388) (industry-report) Forrester analyst recognition of technical debt as mainstream business concern using Q2 2024 survey data, indicating ecosystem maturity and organizational urgency. - **2024-08-21** — [AI Code Search at Mass Scale - Moderne](https://www.moderne.ai/blog/ai-code-search-at-scale-finding-method-invocations-with-natural-language) (case-study) Production deployment of AI-powered semantic code search for large-scale migration across 1,218 repositories and 365K method invocations, demonstrating practical refactoring tooling at enterprise scale. - **2024-08-05** — [Code is 'drowning in security debt' says Veracode – and AI is both problem and solution](https://www.devclass.com/ai-ml/2024/08/05/code-is-drowning-in-security-debt-says-veracode-and-ai-is-both-problem-and-solution/1620841) (industry-report) Veracode analysis of 13 million code scans: 42% of apps have vulnerabilities unresolved >1 year (security debt), with language-specific variance (Java 46% vs. Python 23%). - **2024-07-04** — [The Peril and Promise of AI Writing Code](https://bytewhispersecurity.com/2024/07/04/ai-code-peril-and-promise.html) (opinion) Practitioner field study: AI-assisted null-safety refactoring across 75 files (5000+ changes) succeeds in scope but introduces subtle functional/security issues; efficacy drops sharply with complexity. - **2024-06-11** — [AI-Assisted Coding: Amplifying Bugs and Increasing Developer Risk](https://www.trust-in-soft.com/resources/blogs/ai-assisted-coding-amplifying-bugs-and-increasing-developer-risk) (opinion) Critical assessment documenting AI-generated code increases defect density, security vulnerabilities, and long-term technical debt, highlighting need for integration with robust analysis tools. - **2024-06-03** — [Towards Measuring the Impact of Technical Debt on Lead Time](https://arxiv.org/abs/2406.01578) (research-paper) Industrial case study quantifying TD's mixed impact on development lead time across six components (5-41% variance), showing technical debt metrics alone do not explain deployment delays. - **2024-05-08** — [Refactoring Deep Learning Code: A Study of Practices and Unsatisfied Tool Needs](https://arxiv.org/abs/2405.04861v2) (research-paper) Empirical study of 4,921 refactoring commits across DL projects and 159 practitioner survey revealing current tools inadequately meet practitioner needs for specialized domains. - **2024-04-18** — [Refactoring with AI](https://www.thoughtworks.com/en-in/insights/podcasts/technology-podcasts/refactoring-with-ai) (conference-talk) Thoughtworks industry panel with Martin Fowler and CodeScene CTO citing white paper finding AI-automated refactorings achieved only 37% functional correctness, exposing critical tool maturity gap. - **2024-04-16** — [What Can Self-Admitted Technical Debt Tell Us About Security? A Mixed-Methods Study](https://2024.msrconf.org/details/msr-2024-technical-papers/14/What-Can-Self-Admitted-Technical-Debt-Tell-Us-About-Security-A-Mixed-Methods-Study) (research-paper) MSR 2024 study analyzing 94,455 SATD instances, linking 201 to security CWEs including MITRE Top-25 vulnerabilities, showing critical security dimension of debt detection. - **2024-04-02** — [Automating Technical Debt Management](https://arxiv.org/html/2502.03153v1) (research-paper) MSR analysis of 216 Stack Exchange discussions and 51 TDM tools revealing identification/measurement are top automation opportunities, yet tool errors and poor explainability hinder adoption. - **2024-03-28** — [Generative AI systems, hallucinations, and mounting technical debt](https://dailyai.com/2024/02/generative-ai-systems-hallucinations-and-mounting-technical-debt/) (news-coverage) Coverage of research on technical debt in AI/ML systems (Google paper), warning that ML complexity, hidden feedback loops, and entanglement create 'an avalanche of technical debt waiting to happen.' - **2024-03-27** — [Technical debt's impact on development speed and code quality](https://www.sonarsource.com/blog/technical-debt-s-impact-on-development-speed-and-code-quality/) (industry-report) SonarSource analysis of technical debt trade-offs and their 'Clean as You Code' proactive approach to debt prevention and gradual codebase improvement, reflecting vendor adoption signals. - **2024-03-22** — [Is AI making your code worse?](https://stackoverflow.blog/2024/03/22/is-ai-making-your-code-worse/) (opinion) Stack Overflow discussion with GitClear CEO highlights code quality degradation from AI-generated code: increased churn, reduced readability, and poor test coverage creating long-term refactoring debt. - **2024-03-01** — [Enterprise Generative AI in 2024: The future of work](https://www.altmansolon.com/thought-leadership/2024-enterprise-adoption-generative-ai) (adoption-metric) Survey of 400+ executives shows 78% of software development teams use generative AI tools (up from 23% in 2023), indicating rapid mainstream adoption of AI development tools with security concerns rising. - **2024-02-28** — [Making AI more accurate for automated code refactoring - Moderne](https://www.moderne.ai/blog/ai-assisted-refactoring-in-the-moderne-platform) (product-ga) Moderne Platform integrates LLMs with OpenRewrite's lossless semantic trees for multi-repository refactoring, addressing scalability limits of IDE-based automation for large-scale code improvements. - **2023-12-19** — [Self-Admitted Technical Debt Detection Approaches: A Decade Systematic Review](https://www.arxiv.org/abs/2312.15020) (research-paper) Systematic review of 2014-2024 SATD detection literature showing evolution from NLP to transformer models with improved accuracy, but scalability remains a challenge for industrial deployment. - **2023-12-07** — [Report to the Congressional Defense Committees on National Defense Authorization Act (NDAA) for Fiscal Year 2022 Section 835 Independent Study on Technical Debt in Software-Intensive Systems](https://www.sei.cmu.edu/library/congressional-report-section-835-technical-debt-cmu-sei-2023-tr-003/) (industry-report) CMU SEI report to U.S. Congress documenting that DoD programs are aware of technical debt importance and have established management practices, signaling strategic adoption in defense sector. - **2023-11-16** — [Keyword-labeled self-admitted technical debt and static code analysis have significant relationship but limited overlap](https://oulurepo.oulu.fi/handle/10024/50888) (research-paper) Peer-reviewed study of 33 repositories showing only 15% of self-admitted technical debt comments directly address SonarQube issues, revealing detection tool gaps and complementary detection approaches. - **2023-11-08** — [Tools for Refactoring to Microservices: A Preliminary Usability Report](https://arxiv.org/abs/2311.04798) (research-paper) Review of 22 refactoring tools reveals lack of maturity and generalizability, highlighting adoption barriers even for specialized refactoring scenarios requiring debt remediation automation. - **2023-07-19** — [New Research from Sonar on Cost of Technical Debt](https://securityboulevard.com/2023/07/new-research-from-sonar-on-cost-of-technical-debt/) (adoption-metric) Analysis of over 200 projects quantifying technical debt cost at $306,000 annually per million LOC, providing business case metrics for technical debt management investment. - **2023-07-19** — [Cost of Technical Debt: New Research by Sonar](https://www.sonarsource.com/blog/new-research-from-sonar-on-cost-of-technical-debt/) (adoption-metric) SonarSource analysis of over 200 real-world projects (~11M LOC over 12 months) quantifying technical debt cost at $306,000 per year for a project of one million lines of code, equivalent to 5,500 developer hours of remediation. - **2023-06-28** — [Sonar is destroying my job and it's driving me to despair](https://community.sonarsource.com/t/sonar-is-destroying-my-job-and-its-driving-me-to-despair/92438/7) (opinion) Critical assessment from practitioner showing how static analysis tools misapply technical debt metrics, revealing real-world adoption barriers and tool limitations. - **2023-06-23** — [Drowning in Technical Debt? Uncover Big Risks to Patient Care](https://resources.cerecore.net/drowning-in-technical-debt-uncover-big-risks-to-patient-care) (industry-report) Healthcare IT survey finding 67% of CIOs concerned about technical debt, demonstrating cross-industry awareness of technical debt as an operational risk. - **2023-06-18** — [albertas/deadcode: Find and fix unused Python code](https://github.com/albertas/deadcode) (significant-repo) Open-source Python tool for dead code detection and removal, showing community-driven tooling for technical debt management with 155 GitHub stars. - **2023-06-16** — [Artificial Intelligence for Technical Debt Management in Software Development](http://arxiv.org/abs/2306.10194) (research-paper) Comprehensive academic literature review analyzing 15 research papers on AI techniques for technical debt management, indicating research maturity and academic recognition of the field. - **2023-06-15** — [Automated Code Remediation & the Software Supply Chain](https://www.moderne.ai/blog/free-report-automated-code-remediation-and-the-software-supply-chain) (industry-report) O'Reilly report on automated code remediation for refactoring and securing software supply chains, positioning automated remediation as an industry solution. - **2023-06-02** — [Demystifying the refactoring of machine learning codebases](https://in.pycon.org/cfp/pycon-india-2023/proposals/demystifying-the-refactoring-of-machine-learning-codebases~b4p7b/) (conference-talk) PyCon India 2023 workshop on refactoring ML codebases, showing practitioner-led education and domain-specific refactoring practices. - **2023-06-01** — [Diagnosing Refactoring Dangers](https://arxiv.org/html/2411.08648) (research-paper) Research model and tool for diagnosing refactoring risks and behavior preservation, addressing safety challenges in automated refactoring. ## History - **2026-Sep:** Benchmark evidence hardened the ceiling on autonomous large-scale refactoring: SWE Refactor Bench found only 5.4% of agentic full-repository stack migrations pass all evaluation stages across 520 runs on 8 frontier models (Claude Opus 5 scoring 47/100), and SWE-Bench ProMax found frontier models achieve just 41.2% success on multilingual refactoring versus 50-60% on earlier benchmarks. Production tooling evidence pointed the other direction: Adyen's enterprise-scale OpenRewrite deployment produced 4,000 automated MRs at a 70% merge rate within two months, JetBrains Rider's new refactoring-code skill cut agent task time 83% and cost 63% by invoking IDE refactoring operations directly rather than approximating via text edits, and Sonar's controlled study found cleaner codebases cut AI-agent token consumption 7-8% with 34% fewer file revisits. Countervailing culture-level evidence from GitClear's 4-year, 623M-change longitudinal analysis showed refactoring activity collapsing from 21% (2022) to 3.8% (2026) while copy-paste and code duplication each rose 81%, and a Wizr synthesis of McKinsey/Stripe/Gartner figures reinforced technical debt's growing economic weight (20-40% of technology estate value, 42% of developer time lost weekly). Bounded, human-verified migrations continued to succeed at scale even as autonomous whole-repo refactoring stalled: Mistral documented a named European energy operator's AI-agent migration of 40,000 lines of untested Fortran 77 to C++ with human-verified numerical parity, and Google's Finance Engineering team used Antigravity CLI in headless mode to apply an identical dual-write database migration pattern across 30+ DAOs moving to Cloud Spanner in production. Practitioner and systems-level research converged on a "verification tax" framing: guardrailed prompting (scope lock, required tests, diff-before-edit) raised refactoring accuracy from 40% to 89% in one study, while a broader synthesis found commits up 180% and projects up 50% but releases only up 30%, and production telemetry across 12,400 agent runs showed a 18.2-point gap between benchmark (71%) and real-world (53%) reliability with compounding failures cutting 5-step pipeline success to 36.2% — reinforcing that review and verification capacity, not generation capability, now gate the practice's advancement (cross-analyst synthesis: only 5-8% of enterprises achieve measurable AI ROI at scale). Late-month surveys reinforced the same gap: New Relic (94% rate AI code high quality yet 78% see incident spikes) and Qodo (89% had an AI-related incident, 3.7% call processes sufficient) confirmed post-merge "agent debt", while IBM found nearly 70% of executives expect technical debt to make some AI initiatives untenable, and CodeScene found AI changes fail 60% more often in unhealthy code. - **2026-Aug:** Refactoring ROI gained its first hard economic quantification: a controlled Thoughtworks experiment refactoring an oversized 17,155-line Rust file into 19 smaller files (3,695 lines) cut input token consumption 83% (159,564→27,360), directly pricing code structure under AI-agent workflows. Named deployments reinforced bounded-task success at scale: Visma's 3-million-line .NET modernization achieved 40% effort reduction, NTT DATA reported a 50% faster time-to-market, and a healthcare provider documented $12M in savings with an 85% defect reduction via AI-modernized legacy migration. Countervailing evidence hardened the debt-accumulation case: a synthesis of 2026 industry benchmarks (Cortex, LinearB, CodeRabbit) found PR volume up 20% alongside a 23.5% rise in incidents-per-PR and a 39.9% drop in refactoring activity, with AI-generated code containing 322% more privilege-escalation paths than human code; separately, IEEE's analysis of 304,362 verified AI-authored commits found over 15% introduced quality, security, or maintainability issues. Checksum's State of AI Code 2026 report captured the trust-reality gap directly: 78% of developers trust AI-generated code more than a year ago, yet 61% shipped a production incident within 90 days, 74% had rolled back AI-generated code, and 64.8% said AI code requires more review time than human code. Tooling matured in response: GitLab 19.2 shipped GA agentic automation for dependency remediation and security-backlog review, directly targeting the debt AI-assisted coding has been creating. Later-August evidence sharpened both sides of the split: a fully autonomous $2,430 refactor of a 717k-line TypeScript codebase (189 files, specification-first protocol with 14 refinement and 17 verification cycles) shipped with zero production bugs across 30+ sessions, while loveholidays scaled agent-assisted commits to 80% without sacrificing code health (94% of teams above 9.75 CodeHealth) via CodeScene governance. SWE-Bench ProMax (170 curated multilingual instances) found GPT-5.2 achieves only 41.2% success on production-scale refactoring versus 75%+ on earlier benchmarks, and peer-reviewed research quantified AI's "deletion avoidance" failure mode (29% of models wrap rather than remove dead logic, collapsing SWE-bench success from 63.2% to 41.9% on retrofitted removal tests). Moderne was named a Leader in Gartner's inaugural Magic Quadrant for AI-Augmented Code Modernization Tools, while a multi-case analysis found 8,000+ startups required rescue engineering by mid-2026 from unreviewed AI-generated code, with technical debt up 30-41% post-adoption. Separately, Anthropic's July system-prompt restructuring illustrated a new debt vector: AI tooling platform changes now force downstream customer-side refactoring of prompts, CLAUDE.md files, and eval suites. - **2026-Jul:** A 4-year longitudinal study of 623M code changes (GitClear) quantifies systemic structural decline in the AI era: refactoring activity (moved-lines proxy) down 74%, code duplication up 81%, legacy maintenance down 74%, and error-masking constructs up 47% — documenting write-only mode where generation velocity outpaces refactoring discipline. Peer-reviewed empirical work (26,016 turns, 6 models, 542 tasks) finds 40-73% regression accumulation across multi-turn sessions, with a Verification Gate strategy (retesting new code against prior tests) recovering quality from 75.8% to 87.9%. Meta Engineering confirmed context debt as the refactoring bottleneck: raising agent-usable module context from 5% to 100% via decision records reduced tool calls 40%. New research highlights post-merge maintenance burden: agentic code requires 46% higher corrective maintenance and 45% higher bug-fixing activity post-merge across 182 tracked repositories (one-year longitudinal study). Intent debt emerges as a distinct category—reasoning embedded only in prompts, unrecoverable for maintenance, with 58.8% of prompt-model combinations losing accuracy on model upgrade. Security debt in agent-generated PRs: 38.9% contain security code smells, 82.3% supply chain integrity issues, with 81.1% of hardcoded credentials escaping both automated and human review prior to merge. Further evidence added scale and nuance: a systematic review of 34 empirical studies (2,847 developers) found 58% report code quality degradation and 22.7% technical debt persistence, with early productivity gains (+27% in weeks 1-4) collapsing by week 8 — though mandatory quality gates prevented 67% of debt insertion. A separate analysis of 6,275 mature-codebase repositories found agentic code produces 1.7x more issues per PR, with 110,000+ unresolved issues accumulated by February 2026 and remediation failing to keep pace. Named deployments continued to demonstrate governed success: Datadog used Claude and Cursor for a test-driven production migration of its Stream Router from KV to PostgreSQL, using modularity and parallel infrastructure to enable safe rollout; and a brownfield logistics platform completed 330 merged PRs (~90% AI-generated) over six months by codifying risk zones (80% boilerplate, 20% logic, 0% critical) and documenting context before prompting — reinforcing that bounded, governed brownfield refactoring remains the reproducible success pattern. - **2026-Jun:** Bounded production deployment confirmed alongside new capability benchmarks and maturing governance tooling. Bloomberg's Pomona agentic refactoring tool demonstrated enterprise viability for small-scope autonomous changes: 15/17 PRs merged with <2hr median time-to-close and 80% surveyed engineer adoption interest, showing that human-in-the-loop agentic refactoring generating ~10-line targeted PRs achieves high merge rates in production. CodeScene's Deterministic PR Refactoring Agent reached GA using measurable Code Health signals rather than probabilistic prompting, providing an objective constraint layer for scoping agentic refactoring. However, benchmarking hardened the architectural capability ceiling: SmellBench (294 cases, 7 smell types) showed the strongest agent combination (Qwen Code + Claude Sonnet 4.5) achieving only 50.34/100, with agents failing on cross-file smell detection and architectural understanding; and a practitioner case study confirmed that CI-green, unit-test-passing refactoring still produces production regressions — 7 of 100 Claude Code-refactored Python functions drifted p95 latency from 180ms to 240ms via patterns (double traversal, cache-defeating early returns, asyncpg fast-path breaks) invisible to functional test suites. - **2026-May:** Frontier research establishes hard capability limits for autonomous refactoring, while production deployment validates bounded-task success with governance. Scale AI's SWE Atlas benchmark (70 production refactoring tasks across 10 repos, 6 languages) shows frontier model Claude Opus achieves 48.57% success, with open models lagging significantly and introducing regressions—establishing the lower bound of current agent capability. Peer-reviewed research (arXiv May 2026) quantifies two contrasting phenomena: (1) structured refactoring detection via SBERT + XGBoost achieves F1=0.891 on 339-repo BDD corpus, identifying 692k recurring patterns and cross-org reusable opportunities, demonstrating ML-driven refactoring identification at scale; (2) AI-generated Python refactoring PRs show 22.5% improve quality attributes but 24.17% introduce new violations, with 73.5% merge rate despite trade-offs, documenting the quality-velocity paradox at PR level. Architectural decay research establishes the "Volume-Quality Inverse Law": code volume predicts structural degradation and AI systems produce a "machine signature of defects" invisible to functional correctness testing. SmellBench reveals 63% false positive detection rate on architectural code smell repair, confirming autonomous architectural refactoring remains beyond current LLM agent capability. Production incident prevalence: Censuswide survey (N=500 enterprise IT engineers) shows 89% experienced AI-generated code incidents, 25% suffered complete system outages; 41% report increased manual review time post-AI adoption. METR's landmark randomized trial contradicts productivity perceptions: developers feel 20% speedup but measure 19% slowdown on complex systems; GitClear documents 8x code duplication and 1.57x more security vulnerabilities in AI samples. Named deployment successes persist and validate governance-driven approaches: Novacomp completed Java 8→17 migration of 10k LOC in 50 minutes (vs 3-week estimate) with 60% technical debt reduction and zero regressions; Blue Pearl's Java 11→21 refactoring achieved 90% timeline compression (3 days vs 30+ days) with 92% test coverage; .NET 8 modernization achieved 35% timeline reduction and 60% infrastructure savings. Autonomous agent trials expose context-window failure modes: 14-day Rust API gateway refactoring trial succeeded on warnings and naming but failed at cross-module dependencies after 8 days, accumulating 15 open PRs. Yet practitioner governance innovations demonstrate mitigation: persistent project-memory graph pinned to commits guiding AI consistency; adversarial refactoring workflows using AI for destructive testing rather than generation; mandatory video walkthroughs for AI-generated code changes (12-month platform team case study: 26-55% more code shipped but incident rate +31%, solved via review discipline and observability investment). DORA's ROI framework analysis documents the bounded-task threshold: 35-40% productivity gain on greenfield but only 10% on complex legacy refactoring (Stanford 100k dev study). Late-May practitioner case studies added further production evidence: a 12-month analysis of 4,200 PRs confirmed 26-55% velocity gains alongside a +31% incident rate resolved only through mandatory video walkthroughs, while a 6-month longitudinal peer-reviewed study (95-158 matched engineers) found 84% report productivity gains but 27% report worsened developer experience due to the shift to supervisory work. Systemic governance failure remains measurable: PR review times spike 441% year-over-year; 38% of teams see deployment frequency rise while change failure rate increases in parallel; 41% of AI-generated commits correlate with higher rework rates. The practice remains firmly bleeding-edge: frontier research confirms bounded refactoring (framework upgrades, pattern-based migrations) with engineer oversight succeeds reproducibly at scale, while architectural refactoring, autonomous discovery, and unguided deployment continue creating debt faster than governance frameworks can manage. - **2026-Apr:** Debt accumulation evidence intensified alongside governance signal. Faros telemetry (22K developers, 4K+ teams) confirmed AI as primary code author with code churn +861%, bugs +54%, and incidents +242.7%; BayTech meta-analysis (211M LOC) documented refactoring activity collapsing below 10% of commits while code cloning quadrupled — quantifying the debt creation rate that refactoring practices must now address. KPMG's survey of 2,500 executives found the top 5% of organisations achieve 4.5x ROI through disciplined tech debt governance versus 2x for the average, identifying governance discipline as the key competitive differentiator. A practitioner case study documented Year-2 crisis mechanics precisely: 40% velocity gains in Year 1 reversed into 3.8x maintenance costs by Year 2, resolved only by implementing a governance framework with a 35% AI code cap, 20% sprint debt budget, and tiered review gates. Enterprise case study evidence documented the "throughput trap": teams accumulate surface area faster than they can validate it, with AI building on dead code and ignoring legacy patterns — a distinct organizational governance failure mode separate from individual code quality issues. SonarQube's AI CodeFix reached GA alongside a technical review confirming AI Code Assurance and Quality Gates as key mechanisms for maintaining code quality amid rapid AI-assisted development. OpenRewrite's LST-based recipe engine was validated as foundational tooling for large-scale bounded refactoring (migrations, framework updates, consistency fixes) at scale. SonarQube Remediation Agent and CodeScene's MCP integration (6x more accurate than SonarQube for maintainability prediction) continued to mature. The phase hardened the practice's defining tension: tooling to manage AI-induced debt has matured, but adoption of governance discipline remains sparse — 91% of teams use AI coding but only 15% achieve business value. - **2026-Mar:** Bifurcation deepens with refined measurement evidence. CodeTaste benchmark (arXiv March 2026) quantifies the autonomy-accuracy gap: LLM agents achieve 70% accuracy on specified refactoring tasks but <8% success on autonomous discovery, establishing hard limits on unguided automation. Macro-scale analysis of 8.1M PRs from 4,800 teams documents AI-generated code carries 1.7× higher defects, 30-41% more technical debt, 19% slower delivery despite perceived productivity gains; separately, US technical debt costs are quantified at $2.4T annually with high-debt orgs spending 40% more on maintenance. Large-scale deployment cases prove win-loss split: Airbnb refactored 3.5k React tests in 6 weeks (vs 1.5-year baseline, 75%→97% success); monday.com completed JS monolith breakup in 6 months (vs 8-year baseline) via hybrid AI+engineering approach. CodeScene's agentic refactoring benchmark (Claude Code + CodeHealth MCP) achieved 2–5x code health improvement across 25k files, with Extract Method refactorings increasing 3x under structured guidance—but against an industry baseline of 5.15/10 health vs the 9.4+ threshold required for AI-safe code. Veracode's 150-model security analysis reveals a persistent security ceiling: 55% pass rate regardless of model scale. The practice remains firmly bleeding-edge: structured, supervised refactoring demonstrates concrete productivity gains with oversight, but security debt, correctness failures, and architectural blindness continue to exceed remediation capacity. - **2026-Feb:** Tool ecosystem expansion for managing AI-induced debt accelerates. Moderne extends automated refactoring to Python, signaling market evolution toward practical debt governance as 50% of code changes become AI-generated. Vendor and practitioner evidence validates bounded refactoring success: Holger's Code demonstrates 2-year Delphi migration in 1 week via pattern-based AI execution with human oversight; GitHub Debt Insights deployment shows 45% reduction in debt-related incidents with 3-week advance prediction. Yet the trust-adoption paradox persists and intensifies: SonarSource survey finds 88% of developers report at least one negative AI impact (53% cite unreliable code, 40% duplication), while 93% report at least one positive impact, revealing AI's dual nature. Stack Overflow's ongoing survey shows 84% adoption but only 29% trust, defining trust as willingness to deploy with minimal review—a critical signal that low confidence hinders deployment at scale. Practitioner case studies document nuanced outcomes: Tech Stratos's 40% codebase refactoring showed 30-40% feature velocity gains and 18% test coverage improvement but also critical failures (async error handling, performance regressions, architectural drift), reinforcing that AI amplifies senior judgment but cannot replace it. Verification gap, contextual blindness, and correctness ceiling remain tier-limiting blockers. The practice remains bleeding-edge because while vendor tooling has matured and pockets of successful bounded refactoring exist, the trust-adoption gap signals that organizational readiness for general-purpose AI-driven refactoring has not advanced commensurate with tool capability. - **2026-Jan:** GenAI-induced technical debt becomes measurable and formalized. TechDebt 2026 conference peer-reviewed research quantifies 81 documented cases of GenAI-Induced Self-Admitted Technical Debt (GIST) with specific patterns: verification gaps, incomplete AI-code adaptation, and developer comprehension failures. Moderne's production multi-repo refactoring platform gains enterprise adoption (MEDHOST, Interactions, Allstate, Intel Capital, Choice Hotels) validating vendor maturity. Positive deployment signal: D3 Alpha case study documents successful AI agent refactoring of complex enterprise data systems in weeks vs. prior 76-day manual efforts. Consulting firms (Thoughtworks, BCG, EY, IBM, Xebia) formalize GenAI modernization methodologies with 30-60% efficiency gains on legacy systems. Critical countervailing evidence: Baytech analysis frames rapid AI code generation as creating "Efficiency Paradox" with high-interest debt in final 30% of projects; practitioner guidance emphasizes proactive debt inventory and allocation strategies (20% Rule, Debt Log). The bifurcated landscape persists: bounded refactoring (framework upgrades, legacy migrations) with oversight demonstrates production viability, while unguided large-scale automation continues risk of new debt creation. Verification gap, architectural blindness, and contextual insensitivity remain tier-limiting factors. - **2025-Q4:** Significant production deployment evidence crystallizes the bifurcated landscape. Salesforce publicly documents AI-driven refactoring cutting a 2-year legacy migration to 4 months—a major signal of bounded-task mastery. Yet the adoption-distrust paradox deepens: Stack Overflow's year-end survey (49K+ developers) shows 80% adoption paired with trust collapsing to 29%, with 66% reporting extra time fixing AI-generated code. JetBrains' global ecosystem survey (24.5K developers) confirms 85% regular AI use and 62% reliance on AI coding assistants, but developers express skepticism about productivity metrics in technical debt contexts. Practitioner case studies reveal nuanced deployment: Altom's test automation refactoring achieved acceleration with AI but required human expertise for complex inheritance patterns. Developer communities document critical refactoring failures: over-application of DRY principles creating "Conditional Monsters," wrong abstractions, and silent state-handling bugs. Security dimension intensifies: real breaches (Toyota, Decathlon) trace to legacy migration and misconfiguration debt from rapid AI-assisted deployments, underscoring data risk consequences. The practice remains bleeding-edge: bounded refactoring (legacy migrations, test suite modernization, pattern-based transformations) achieves repeatable success with engineer oversight, but unguided large-scale automation continues to generate technical debt, and the trust-adoption gap signals organizational barriers persist unresolved. Correctness, architectural understanding, and contextual sensitivity remain tier-limiting factors preventing production-ready general-purpose AI refactoring. - **2025-Q3:** Real-world deployment evidence emerges with qualified success: Cognizant demonstrates 35% cost and 25% effort reduction on major Java migrations; European telecom leader successfully modernizes monolithic legacy architecture via phased, engineer-guided AI refactoring. Simultaneously, GitClear's analysis of 211M LOC confirms refactoring has become the weakest link in the development cycle—2025 marks the first year code duplication introduction exceeds refactoring activity, indicating AI-generated code creates technical debt faster than automated remediation can address it. Stack Overflow's 49K-developer survey (July 2025) and Google DORA research (~5K professionals) both show persistent trust deficit: adoption remains near-universal but practitioners remain reluctant. Fastly's developer survey documents that 33% of senior developers ship >50% AI-generated code (vs. 13% junior), signaling differentiated deployment by experience level. Academic research (IEEE) on distributed system refactoring reveals failure propagation risks across service boundaries, underlining the context-sensitivity of safe refactoring. The tier-limiting tension persists: bounded refactoring tasks (major framework upgrades, null-safety migrations, standardized patterns) show measurable success in controlled deployments, but the paradox deepens—automation addresses only the small subset of refactoring work where AI correctness is highest, while the bulk of technical debt remains in legacy, novel, and domain-specific code where AI-assisted approaches continue to generate new debt. - **2025-Q2:** Web developers report 61% of AI-generated code requires refactoring due to readability and duplication issues, confirming technical debt creation at deployment scale. SonarQube's AI Code Assurance adoption grows among federal agencies seeking compliant AI validation workflows. Practitioner experiments document critical refactoring failures: type mismatches breaking production code, architectural misunderstandings in autonomous transformations, silent state-handling bugs. Survey data shows 82% of developers using AI assistants daily but two-thirds report AI missing critical context for large refactoring tasks. Enterprise tooling demonstrates bounded-task success (40% lead time improvements from complexity reduction via targeted debt fixes) but AI-generated code carries 1.7x higher defect density than human code. Verification gap and contextual blindness remain tier-limiting: deployment-scale refactoring automation achieves success only in constrained domains, while general-purpose production refactoring remains unresolved. - **2025-Q1:** AI tool adoption reaches 98% of developers (near-universal), but verification practices stagnate—96% don't fully trust output, only 48% verify before commit. Technical debt created by AI-generated code accelerates sharply: 8-fold increase in code duplication, 10x redundancy since 2022, 7.2% delivery stability decline. Vendor tooling pivots to remediation: SonarQube releases AI Code Assurance and AI CodeFix to validate and auto-fix AI-generated code. Accenture quantifies US technical debt at $2.41 trillion annually. Practitioner analysis (GitLab, Tomassetti, RecodeX) documents persistent limitations: AI excels at standard patterns but fails on symbol resolution, context-dependent logic, and legacy systems. The practice remains bleeding-edge: verification gap, correctness ceiling, and organizational trust barriers prevent general-purpose automated refactoring from reaching production maturity despite pockets of success (bounded task automation, young codebases). - **2024-Q4:** AI tool adoption reaches 84% of developers (up from 76% in Q3), but trust plummets to 46% distrust (from 31%), signaling adoption-confidence divergence. Refactoring correctness ceiling remains: ChatGPT achieves 63.6% expert-parity rates, Gemini 56.2%; legacy tools below 50%. New concern: AI-generated code creates technical debt faster than manual development (CAST November 2024). Vendor consolidation toward "compliance-grade" refactoring (Byteable, CAST Highlight) reflects market shift to risk management. Academic research (Journal of Systems and Software, December 2024) surveys technical debt in AI-enabled systems; arXiv papers on IDE trust safeguards signal focus on adoption barriers. Practitioner consensus emerges: LLMs excel in bounded domains (null-safety, standard patterns, young codebases) but worsen debt in legacy/novel code. Correctness and organizational trust remain tier-limiting factors. - **2024-Q3:** Correctness crisis deepens—extended analysis reveals Copilot 46.3%, ChatGPT 65.2%, CodeWhisperer 31.1% correctness rates; code churn doubling by 2024; 80% of enterprises report technical debt stifles innovation. Security debt quantified at scale: Veracode's 13M scans show 42% of apps have >1-year-old unresolved vulnerabilities (Java 46%, Python 23%). Developer trust in AI coding tools remains low (42% trust output, 45% report inadequacy for complex tasks despite 76% adoption). Positive signal: Moderne demonstrates production-scale AI semantic code search across 1,218 repos; practitioner successes with bounded refactoring tasks (null-safety migrations, 5000+ changes). BERT models emerge as most effective for technical debt detection (arXiv September 2024). Organizational and technical barriers persist as tier-limiting factors. - **2024-Q2:** AI-assisted refactoring correctness crisis exposed—industry analysis reveals only 37% of AI-generated refactorings are functionally correct in production contexts, setting back automation confidence. Domain-specific studies confirm existing tools inadequately meet practitioner needs (e.g., deep learning projects). Security dimension of technical debt clarifies: SATD detection research links debt comments to MITRE Top-25 vulnerabilities, establishing debt as a security concern. Measurement research shows technical debt's impact on velocity is inconsistent and context-dependent; identification and measurement emerge as the most sought-after automation activities, yet tool adoption faces barriers from explainability gaps. - **2024-Q1:** Generative AI adoption in software development reaches 78% (up from 23% in 2023). Vendor landscape accelerates: Moderne integrates LLMs with OpenRewrite for multi-repo refactoring. Simultaneously, evidence emerges that AI-generated code creates new technical debt (reduced readability, poor test coverage, code churn). Research warns ML systems themselves accumulate structural debt faster than manual development. Tool fragmentation persists; organizational tension between metrics-driven and developer-perceived debt remains unresolved. - **2023-H2:** Strategic adoption in defense sector formalized (DoD programs establish management practices per CMU SEI report); academic research maturation shows evolution to transformer-based SATD detection; industry metrics quantify financial impact ($306K per million LOC per year); tool maturity assessments reveal production readiness gaps even for specialized refactoring. Organizational barriers persist: static analysis tools show only 15% overlap with developer-identified debt, revealing need for complementary detection approaches. - **2023-H1:** Academic research on AI for technical debt management emerging (literature review, safety models); vendors (Moderne) positioning automated remediation; community tooling for dead code detection visible but limited adoption. Key blocker: tool interpretation and organizational understanding of debt as a structural rather than metric-driven problem. ## Tools - [CodeScene](https://codescene.com) - [OpenRewrite](https://docs.openrewrite.org) - [Moderne](https://www.moderne.io) - [JetBrains Rider](https://www.jetbrains.com/rider/) - [SonarQube](https://www.sonarsource.com/products/sonarqube/) _Source: https://www.thestateofplay.ai/practice/code-refactoring-and-technical-debt-management — CC BY 4.0._