The AI landscape doesn't move in one direction — it lurches. Some techniques leap from experiment to table stakes in a single quarter; others stall against regulatory walls, technical ceilings, or organisational inertia that no amount of hype can dislodge. Knowing which is which is the hard part. The State of Play cuts through the noise with a rigorously maintained index of AI techniques across every major business domain — classified by maturity, evidenced by real-world adoption, and updated daily so you always know where you stand relative to the field. Stop guessing. Start knowing.
A daily newsletter distilling the past two weeks of movement in a domain or two — delivered to your inbox while the index updates in the background.
Each dot marks the weighted maturity of practices within a domain — hover for a brief summary, click for more detail
AI that drafts performance reviews and synthesises 360-degree feedback into actionable insights for managers and employees. Includes review narrative generation and multi-source feedback analysis; distinct from engagement analytics which measures organisational sentiment rather than individual performance.
AI-powered performance review generation and 360-degree feedback synthesis has proven its value at forward-leaning organisations but remains far from mainstream adoption. The tooling works: named deployments show halved drafting time, sharply higher completion rates, and measurable attrition improvements. A handful of platform vendors and dedicated specialists now offer production-grade capabilities, and the 360-feedback market is projected to roughly double by 2033. Yet the practice is stalled at leading-edge maturity, held back less by technology than by everything surrounding it. Most enterprises testing generative AI in HR never reach production. Eighty percent of users abandon AI tools within weeks, a failure of change management rather than capability. Regulatory exposure is tightening, with California's Automated Decision Systems rules now imposing vendor liability and peer-reviewed research finding that algorithmic evaluation feels less respectful than human evaluation to the people being assessed. The deeper structural question is whether AI synthesis can rescue a review format that few HR professionals consider effective. The market has bifurcated accordingly: premium bias-aware platforms on one side, commodity LLM-based tools on the other, with organisational readiness -- not feature completeness -- as the binding constraint on further adoption.
Mainstream adoption is now documented at scale across major organisations: a May 2026 survey of 1,300+ managers shows 91% use AI in performance assessment, and Meta, Google, and JPMorgan have formally tied AI usage to performance reviews and compensation decisions. Lattice, Culture Amp, and Textio anchor the production vendor ecosystem, with Culture Amp's AI Coach (GA March 2026) synthesizing performance data across 1.5B+ data points, and Workday's Talent Management Agent (GA March 2026) drafting evidence-based reviews embedded directly into enterprise workflows. Seventy percent of talent management executives expect increased AI use in performance review development within one year (SHRM). Market analysis projects 360 feedback software growing from USD 1.5B (2024) to USD 3.2B (2033) at 9.5% CAGR, with new entrants expanding market tiers: KS-Agents (€49/mo SMB tier, EU AI Act compliant), TalentGuard (3M+ assessments, 400+ customers, 98% completion), and Anymize (production privacy-preserving 360 synthesis with 2–3 hours per manager consolidation). Productivity claims dominate vendor messaging, yet execution remains uneven. Citi's Performance Assist, JPMorgan's LLM Suite, and BCG's internal AI assistant all report 40% reduction in review-writing time, but Harvard Business Review analysis identifies critical upstream data issues: models amplify incomplete manager visibility (bias is baked into which projects managers saw, which communications were included in metadata feeds) and generate polished versions of already-biased assessment patterns.
Yet the perception-reality gap on productivity undermines mainstream adoption claims. METR's 2026 controlled trial of 16 experienced developers shows AI tools increased task completion time by 19%, despite developers self-reporting 2x value increase and 3x speed gains before starting. Harness survey of 700 engineers found code review time increased 81% due to overhead of reviewing generated code, with 89% of engineering leaders believing their metrics accurately capture AI impact—revealing fundamental measurement failures across the ecosystem. Adoption barriers persist despite platform maturity: continuous feedback models face 7-month survey fatigue degrading data quality; pricing friction and integration complexity drive platform abandonment; employee sentiment remains mixed (33% positive, 35% negative on AI involvement in reviews). Real-world failures surface even at sophisticated organisations: Mastercard's attempt to automate 360-degree reviews using Copilot failed due to tool confusion and generation of identical feedback across individuals. Organisations succeeding with these tools invest heavily in governance frameworks (emerging Data Maturity Matters standard for people data audit trails) and human review processes—emphasizing that AI serves as drafting support, not decision automation. Regulatory environment tightens further: EU AI Act (effective August 2026) classifies performance reviews as high-risk with human oversight requirements, and Mobley v. Workday class action extends discrimination litigation directly into performance management algorithms. The binding constraint has shifted from technology capability to organisational readiness: deployment proven, ROI claimed but not consistently measured, but implementation complexity and measurement discipline remain primary barriers to wider adoption.
As of July 2026, vendor ecosystem advancement continues: Lattice launches Model Context Protocol (MCP) integrations enabling performance review drafting directly within Claude, ChatGPT, and Slack—the first HR platform to officially land in frontier AI marketplaces—while also releasing AI Agent for 1:1s that synthesizes conversation data into evidence-based review drafts. Named deployments show strong adoption (Smartcat achieved 100% participation in first company-wide review cycle across 38 countries), yet a critical implementation gap surfaces: executives report 89% satisfaction with AI-transformed reviews while employees report only 40%—a 6× perception divergence reflecting incomplete change management and low shared understanding of organisational AI vision. Governance frameworks mature alongside capability: HiBob and emerging standards now operationalise risk management against bias, over-automation, and hallucination risks using NIST structures applicable to Title VII compliance and August 2026 EU AI Act requirements. The core constraint remains unchanged: technology is production-ready, but organisational readiness—encompassing measurement discipline, governance infrastructure, change management capability, and employee trust—determines further advancement.
— SHRM survey (1,908 HR professionals, Dec 2025) shows 39% with AI in HR, 62% using AI somewhere, but 56% do not formally measure AI success—documents adoption breadth with critical measurement discipline gap.
— Lattice GA of Evidence-based Review Drafts pulling from 1:1s, goals, and feedback; early customers completed reviews 47% faster with human oversight preserved—demonstrates vendor product advancement in evidence-grounded synthesis.
— PYX Labs benchmark across 7 frontier models on 84 employee-listening tasks reveals synthesis scores as low as 33% vs 64–82% for quantitative tasks; Resume Builder survey shows 60% of managers use AI for people decisions despite capability gap—documents critical adoption risk.
— Peer-reviewed experiment (N=192) finds fully AI-generated feedback produces less favorable responses than human feedback; hybrid human-AI models perform comparably to human-only—reveals adoption barrier despite employee acceptance of hybrids.
— Teachers College Columbia deploys Lattice AI-supported performance management (ACE platform) launching September 1, 2026; independent educational institution selecting platform for feedback synthesis and review support.
— Culture Amp CPO documents adoption paradox; 84% confidence in AI post-training, but real behavior unchanged on normal workdays—reveals identity/unlearning barriers prevent task-level adoption, not technical capability.
— Analysis of 74% pilot-to-production stall rate despite $301B global AI spend; root causes—identity threat, capability gaps, performance incentive misalignment, workflow control anxiety—directly explain why performance review AI remains leading-edge despite tool maturity.
— Culture Amp analysis of 98% AI usage with 2-3× higher transformation at agentic deployment level (automation + workflow orchestration) vs task-level use. Named practitioner examples (Sarah Ostrom, Kate Sargent) document real-world bounded autonomy implementations in performance workflows.
2023-H1: Early production deployments at mid-market (Lattice platform) with high engagement completion rates. Tech giants (Google) experimenting with AI integration into review inputs. Effectiveness concerns persist; meta-analysis suggests 360 feedback yields limited performance impact.
2023-H2: Production deployment of 360-synthesis in HR platforms (Stryve case study, Lattice TrustRadius adoption). Generative AI pilots for feedback summarization (Culture Amp). Return to structured review cycles (50% of companies). Regulatory barriers emerge (EEOC Title VII guidance on AI discrimination). Critical research reinforces limitations of 360-degree feedback design independent of AI.
2024-Q1: Lattice and Culture Amp continue platform innovation with AI-powered engagement insights and review analytics. Forrester TEI study shows 311% ROI from performance management platforms, driving enterprise adoption. Cautionary signals intensify: vendors (Textio) document gender bias in feedback generation (22% more personality feedback for women); legal guidance (National Law Review, NYC AI law) establishes compliance risks; practitioners warn against general-purpose LLMs for reviews. Adoption momentum remains steady but constrained by discrimination liability concerns and tool limitations.
2024-Q2: Major platform feature launches accelerate feedback synthesis: Lattice adds AI Feedback Summaries and AI Engagement Insights; Culture Amp deploys AI Comment Summaries across 400+ customers (6,600 hours saved, 70% satisfaction). Named deployment case study (Velera) shows 67% quality improvement and 50% time reduction with purpose-built AI. Platforms diverge into premium (bias-aware) and commodity (general-purpose LLM) tiers; regulatory and effectiveness barriers persist.
2024-Q3: Dedicated vendor ecosystem matures: Textio launches performance review AI with bias mitigation (July). Lattice reports 40% customer activation of AI engagement insights within weeks of launch. Broad GenAI adoption accelerates (24% of U.S. workers using GenAI weekly by August). Regulatory environment tightens sharply: federal and state employment law explicitly classify performance reviews as "selection procedures" with Title VII discrimination liability. Market consolidates into premium (bias-aware) and commodity (general-purpose LLM) tiers; structural 360 feedback limitations persist independent of AI quality.
2024-Q4: Platform maturation continues: Lattice (October) showcases AI assistant and HRIS with named customer deployments at Lattiverse; Culture Amp serves 6,500 companies with bias-mitigation synthesis. Workforce sentiment shifts cautiously positive (SAP survey: 55% support AI integration into reviews, 64% among high-AI-literacy workers). GenAI workplace adoption broadens (75% of enterprises, EY survey across 23 countries). Trust barriers persist sharply (Thinkers360: 75%+ concerned about AI trustworthiness, 80% on bias). Structural limits resurface: industry consolidates toward continuous feedback models (Accenture 2016, Yahoo 2022 abandoned annual reviews), suggesting AI synthesis cannot salvage declining effectiveness of traditional annual cycles themselves.
2025-Q1: Platform adoption accelerates with AI-powered conversational feedback gaining traction (85% Fortune 500 usage, 90% leader satisfaction). Quantified business case emerges: McKinsey/PwC/Deloitte research documents 40% productivity gains, 30% bias reduction, 15% engagement lifts. Yet fundamental adoption barriers intensify. Peer-reviewed research (UC Berkeley Haas, 995 participants) finds algorithmic evaluation produces lower perceived respect than human evaluation—a barrier independent of bias. Industry experts (Korn Ferry) warn AI overreliance risks data inaccuracy and loss of human touch, particularly given only 2% of HR professionals believe traditional performance reviews are effective. Market bifurcates: premium bias-aware solutions (Textio, Culture Amp, Lattice) vs. commodity general-purpose LLM tools, with regulatory liability concerns constraining enterprise adoption.
2025-Q2: Platform vendor momentum continues: Lattice enhances AI Writing Assist and launches always-on AI Agent (May), Culture Amp refreshes Perform with multilingual synthesis (May), Textio expands to 200K Chrome extension users (June). Yet legal barriers sharply intensify: Supreme Court's Muldrow ruling (April) lowers bias litigation threshold; employment law experts warn of Title VII cascades from algorithmic amplification. Implementation failures emerge: case studies show AI monitoring boosting productivity 25% but doubling turnover within six months. Critical expert commentary (David Ferrucci, Fortune June) articulates surveillance and creativity-stifling risks. Practice remains locked at leading-edge maturity but constrained by legal exposure, implementation failure modes, and structural limitations of annual review cycles themselves.
2025-Q3: Vendor GA momentum continues: Culture Amp releases AI Coach for feedback conversations (July), expanding feedback synthesis tooling across Engage and Perform platforms. Textio maintains 25% Fortune 500 penetration with Lift product. Market analysis (Gallup, Sprad) confirms 12.5% productivity boost from 360 feedback, with market growing toward $2.5B by 2032. Yet regulatory barriers sharpen further: California Automated Decision Systems regulations (effective October 1, 2025) establish vendor liability and algorithmic non-discrimination requirements for performance AI. Practitioners identify persistent tooling gaps: generic HR platforms (Lattice, Workday) lack work-based data integration and remain time-intensive for engineering-focused reviews, exposing limitations of commodity-tier solutions. Practice remains locked at leading-edge, constrained by regulatory tightening, implementation complexity, and market bifurcation between premium bias-aware and commodity LLM-based tools.
2025-Q4: Conversational feedback AI emerges: Culture Amp advances AI Coach into feedback conversations, signaling market shift beyond structured review synthesis. Named mid-market deployments show concrete ROI: Figma reduced review drafting 50% (1hr to 30min), GoCardless achieved 100% review completion (vs 62%), Vantage West reduced attrition 27%. Adoption sentiment shifts mainstream: 40% of HR leaders rank performance management top priority for 2026, 42% regularly using agentic AI (up from 24% workers in Aug 2024), 83% optimistic about AI in HR despite 61% ethical concerns. Yet legal and structural barriers intensify sharply: Mobley v. Workday (filed May 2025, conditional class certification Oct 2025) extends age/race/disability discrimination litigation to performance management; regulatory enforcement targets AI bias and transparency (Colorado, OMB M-26-04); industry research documents 60% of performance issues stem from systemic/cultural factors, not individual capability—constraining impact of individual-level feedback synthesis. Practice locked at leading-edge: deployment proven, ROI documented, but regulatory liability, legal exposure, and structural limitations constrain further tier advancement.
2026-Jan: Market transitions from experimentation to results focus, yet deployment barriers dominate. Industry forecasts shift (Workday) to end of AI experimentation phase with emphasis on practical, orchestrated AI systems in HR. Real-world adoption data (Databricks/Economist) shows stark gaps: 85% of enterprises test GenAI but only 22% confident in IT architecture; 60% UK cases still not in production. Production failure analysis (HyperSense/RAND/Gartner) documents 88% of AI agent projects never reaching production due to data fragmentation and integration barriers. User adoption (Microsoft) shows 80% tool abandonment within 3 weeks, driven by organizational readiness gaps, not tool quality. Real deployments remain positive (Big Light case study: 70% participation, emphasis on human judgment in synthesis). Yet HBR guidance reinforces: 360 effectiveness requires dialogue and human discussion—automation alone insufficient. Emphasis shifts from capability to implementation sustainability as dominant constraint.
2026-Feb: Platform momentum accelerates with feature proliferation. Lattice introduces AI Review Summaries for performance calibrations and AI agents for HR automation; Teamspective enters market with AI-integrated platform for performance insights. Market data (Sopact) projects 360-feedback market growth from $1.16B to $2.27B by 2033, yet growth constrained by deployment barriers: 80%+ of AI agent projects fail to reach production; 80% of users abandon AI tools within 3 weeks despite capability. Deployment barriers crystallize as primary constraint—not tool maturity, but integration complexity, data fragmentation, and organizational readiness. Practice locked at leading-edge: deployment proven, ROI demonstrated, but implementation barriers prevent further advancement.
2026-Mar: Regulatory landscape sharply tightens while vendor ecosystem expands. EU AI Act (effective Aug 2026) classifies performance review systems as high-risk with €35M penalties, requiring human oversight and transparency; state-level US regulations (NYC, Colorado, Texas, Illinois, Maryland) impose mandatory bias audits and liability on vendors. Adoption data shows 13% of employers using AI in performance reviews (SHRM survey), with 90% of employees believing AI fairer than manager-written reviews—signaling growing employee appetite tempered by implementation caution. New product innovation: Your360 AI launches voice-powered 360 feedback in GA after 12+ pilot deployments (1,000+ conversations), pricing at $2,500 for 8 participants against a historical $5-15K benchmark; Workday deploys Sana agentic AI agents for performance review prep. Vendor ecosystem matures: AI now standard across Lattice, Culture Amp, 15Five, indicating table-stakes functionality. However, organizational readiness remains primary barrier—mainstream guidance emphasizes human-centered AI supporting decisions rather than pure automation. Practice remains locked at leading-edge: technology proven, adoption growing, but regulatory tightening, implementation complexity, and structural limitations of annual reviews constrain advancement.
2026-Apr: Mainstream adoption is documented at scale — survey of 1,300+ managers found 91% use AI in performance assessment, with Culture Amp's AI Coach (1.5B+ data points) and Workday's Talent Management Agent both reaching GA in March 2026. Market analysis confirms ecosystem maturation: 360 feedback software projected to grow USD 1.5B (2024) to USD 3.2B (2033) at 9.5% CAGR. New entrants expand market tier: KS-Agents (€49/mo SMB tier, EU AI Act compliant), TalentGuard (3M+ assessments, 400+ customers, 98% completion), Atlas Cowork (live CRM/Jira data integration for bias-aware synthesis). Yet output quality remains material constraint: independent hands-on testing (ToolsRadar 6-week evaluation) shows Lattice AI features are LLM-generated suggestions with integration failures when reading stale data; Workday research shows 40% of AI time savings lost to fixing low-quality outputs. Adoption barriers extend beyond technology: continuous feedback models face 7-month survey fatigue degrading data quality; pricing friction and integration complexity drive platform abandonment. Market consolidation signals shift: Lattice discontinued HRIS to focus performance management as defensible product line. Mastercard's Copilot-driven 360 automation failed due to identical feedback and ignored constraints—underscoring that deployment breadth has outpaced reliability.
2026-May: Perception-reality gap in productivity gains becomes visible. METR controlled trial documents 19% slowdown in task completion despite self-reported 2x gains; Harness survey shows code review time increased 81% due to AI output overhead. Meta, Google, and JPMorgan formally tie AI usage to performance reviews and compensation but lack demonstrated ROI. HBR analysis documents upstream bias amplification—models polish existing assessment blind spots rather than correcting them. Governance frameworks emerge (Data Maturity Matters standard for people data audit trails) signaling requirement for data provenance and appeals processes. Employee sentiment mixed (33% positive, 35% negative) revealing organizational readiness gap. Seventy percent of talent executives expect increased AI in performance review development (SHRM), yet implementation discipline and measurement capability remain binding constraints on further adoption. Production 360-feedback solutions expand (Anymize privacy-preserving consolidation; 10–20% response rate improvement from anonymity guarantees).
2026-Jun: Market maturation combined with critical adoption barriers crystallizes leading-edge plateau. Mordor Intelligence projects 360-feedback market growth from $1.35B (2026) to $2.16B (2031) at 9.87% CAGR, with 48% of European HR teams piloting specialized AI tools—confirming mainstream adoption trajectory. New platform entrants and HRIS consolidation signal table-stakes AI: HiBob's Bob HRIS embeds AI-powered performance review summaries for multi-rater feedback synthesis directly into unified data model; Sprad's Talent Management Workspace integrates reviews, 360 feedback, goals, and AI agents into daily workflow rhythm. Adoption metrics solidify: 47% of organizations now factor AI-generated feedback into formal performance evaluations (General Assembly), with average 4-hour manager time savings per review cycle via AI consolidation; 59% of UK managers now use generative AI to help write reviews (Visier research), while Teamflect's analysis of 11,000 reviews found 61% of quality is driven by narrative comments rather than scores—precisely the gap AI synthesis addresses. Vendor momentum continues with June announcements: Lattice (Lattiverse June 10) launches evidence-based review drafts from real 1:1 data, MCP integration, and AI agents for coaching; Koji distinguishes 360 tools with AI-moderated follow-up conversations converting Likert ratings into behavioral stories. Yet measurement crisis deepens: two-thirds of enterprises rely on estimates rather than measured results for AI ROI; MIT NANDA and Gartner data document 88% of AI automation projects abandoned before full deployment and 40% of agentic AI projects predicted cancelled by 2027, compounding the organizational readiness gap. Negative adoption signals emerge: SHRM survey shows only 17% describe AI implementation highly successful, only 14% use AI in core systems; specific deployments at global financial services (14% coaching increase, 3-point fairness improvement, 9% attrition reduction against control) demonstrate real impact but remain outliers. Critical research reveals adoption ceiling: controlled study (242 managers) documents only 42% compliance with algorithm low-rating recommendations vs 60% for high ratings; managers systematically override low scores citing relationship concerns—leniency bias persists as fundamental behavioral barrier. Governance frameworks emerging: EU AI Act (effective August 2, 2026) classifies performance evaluation as high-risk with €15M/3% turnover penalties; Data Maturity Matters standard establishes audit trail and appeal process requirements. Implementation failures documented: mid-sized logistics company's AI over-weighted response-time metric, flagging high performers as underperformers; Mastercard's Copilot-driven 360 automation generated identical feedback across individuals. Practice remains locked at leading-edge: technology mature and adopted at scale, but fundamental barriers intensify—measurement discipline, governance complexity, legal liability, manager behavioral resistance, and unproven ROI preventing mainstream advancement.
2026-Jul: Lattice became the first HR platform to ship MCP integrations into ChatGPT, Claude, and Slack, and launched an AI Agent that synthesizes 1:1 conversations into Evidence-based Review Drafts—marking a shift toward frontier-model-native review generation. A BetterWorks survey exposed a stark perception gap (89% of executives satisfied with AI-augmented reviews vs. only 40% of employees, a 6x divergence), reinforcing adoption-barrier evidence that just under 30% of organizations consider AI performance-management deployment broadly successful despite 78% usage. Culture Amp's 2026 AI in HR Study (264 HR professionals) found 77% buy-in but only 24% comfort with agentic autonomy — down from 86% confidence the prior year — while a companion analysis showed agentic-level deployment drives 2-3x higher transformation than task-level use. A PYX Labs benchmark of 7 frontier models across 84 employee-listening tasks found interpretive synthesis (core to 360 feedback) scoring as low as 33% versus 64-82% for quantitative tasks, and a documented failure at Meta showed activity-based scoring misclassifying employees on protected FMLA/ADA leave as zero-performers.
2026-Aug: SHRM's 2026 State of AI in HR report (1,908 professionals) confirms 39% formal AI-in-HR adoption and 62% usage overall, but 56% still do not formally measure AI success. Lattice reached GA with Evidence-based Review Drafts pulling from 1:1s, goals, and feedback, with early customers completing reviews 47% faster; Teachers College Columbia adopted the Lattice-powered ACE platform for a September 2026 launch. A PYX Labs I-O benchmark reinforces the interpretive-synthesis gap (33% vs. 64-82% for quantitative tasks) against a Resume Builder finding that 60% of managers already use AI for raise and layoff decisions, and a peer-reviewed experiment (N=192) found fully AI-generated feedback rated less favorably than human or hybrid feedback. Culture Amp's CPO described an adoption paradox—84% post-training confidence with unchanged day-to-day behavior—underscoring that identity and habit, not capability, remain the binding constraint.