Perly Consulting │ Beck Eco

The State of Play

A living index of AI adoption across industries — where established practice meets the bleeding edge
UPDATED DAILY

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.

The Daily Dispatch

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.

AI Maturity by Domain

Each dot marks the weighted maturity of practices within a domain — hover for a brief summary, click for more detail

DOMAIN
BLEEDING EDGEESTABLISHED

Trend identification & horizon scanning

LEADING EDGE

TRAJECTORY

Stalled

AI that identifies emerging trends and weak signals across large volumes of publications, filings, and discussions. Includes early signal detection and trend trajectory modelling; distinct from social listening which monitors social platforms rather than scanning broad information sources.

OVERVIEW

AI-driven horizon scanning has proven its value in specific verticals -- IP intelligence, regulatory monitoring, research foresight -- but remains stuck at the leading edge, unable to break into broader corporate adoption. The tools work: production systems scan millions of sources, detect weak signals, and model trend trajectories with measurable efficiency gains. What has stalled is not capability but trust. A growing share of the information environment is AI-generated, degrading the signal-to-noise ratio that these systems fundamentally depend on. Combined with persistent governance gaps and high implementation costs, the result is a practice bifurcated between thriving vertical deployments and blocked horizontal expansion. The core tension is architectural: horizon scanning tools have matured faster than the information ecosystem they scan.

CURRENT LANDSCAPE

Vertical deployments continue to show genuine production maturity across IP, regulatory, consumer intelligence, defense, and research domains. Patsnap's IP platform serves 18,000+ users with 75% faster innovation cycles and 25% lower R&D costs; the March 2026 Pulse feature release extends continuous trend monitoring capabilities. Regulatory scanning via LEGALFLY and other vendors achieves 50% time reduction in compliance monitoring. CPG trend identification—exemplified by i-Genie's deployment enabling $70M incremental revenue by identifying consumer trends 4-6 months ahead of competitors—demonstrates near-real-time trend identification translating directly to revenue impact. UK Defence Science and Technology Laboratory deployed AI-powered horizon scanning in April 2026, processing 300K+ articles monthly with analyst hit rates improving from 1% to 40%, winning the 2025 Analysis in Government Award; the European Medicines Agency embedded horizon scanning as strategic capability for regulatory preparedness. AI adoption in the IP ecosystem has reached 85%, up from 57% in 2023. Government and institutional use is expanding: UK Defra deployed horizon scanning via Futures Toolkit, EU's FUTURINNOV project runs systematic foresight exercises, OECD published policy guidance on building horizon scanning capacity (covering 10,000+ technology signals), and the UK Office for Science conducted government-scale surveys for emerging technology assessment. Large-scale platforms (TSC.ai, SAI360, Horizon Scan AI) operate at enterprise scale across 100+ countries, with Horizon Scan AI deployed across UNDP, UNDRR, OECD, and the European Commission. Institutional-scale deployments expanded in late April: German Bundestag's Technology Assessment Office (TAB) has operated systematic horizon scanning since 2014 combining software analysis with expert validation; the European Union's ENISA (cybersecurity agency) published comprehensive signal identification methodology; major think tanks including CSIS and the Atlantic Council operationalize horizon scanning with expert surveys (Global Foresight 2036 capturing 450+ geostrategist forecasts); and corporate programs at firms like G+D institutionalized trend monitoring methodologies. Agentic AI tools now reach production: Silent Eight's Horizon Scanning Agent provides continuous regulatory and geopolitical monitoring with governance guardrails and human-in-the-loop decision support.

Methodological maturation is accelerating alongside growing recognition of AI quality limitations. The Journal of Medical Internet Research (April 2026) published the first standardized 35-item reporting checklist for horizon scanning studies, addressing reproducibility and field-wide comparability. Thoughtworks' April 2026 Technology Radar identified a macro trend shift toward "harness engineering"—infrastructure, constraints, and feedback loops designed to improve AI agent reliability—marking a transition from experimental AI to production reliability focus. Analyst firms continue identifying market-scale trend signals: Stanford's 2026 AI Index documents dual trends (capability breakthroughs but 45% spike in AI misinformation), and Forrester's April 2026 predictions identify market correction dynamics (25% of planned AI spend deferred to 2027, only one-third of leaders tie AI to financial outcomes). However, empirical research published in Harvard Business Review found that leading LLMs converge heavily toward culturally fashionable recommendations when asked for strategic advice, producing "trendslop" rather than contextual analysis—a critical signal that AI-augmented trend identification systems require architectural safeguards against bias toward consensus narratives. Production tools operationalize the practice: Claude Code published a skill for automated trend identification with weak signal detection, adoption curve analysis, and trend classification frameworks; KHネオケム deployed PatSnap Eureka agents for multi-dimensional business trend analysis.

These successes, however, have not translated into broad corporate foresight adoption. The barrier is increasingly environmental rather than technical. Analysis of early 2026 data suggests roughly 40% of web content is now AI-generated, contaminating the weak-signal streams that horizon scanning fundamentally depends on. Latest-generation reasoning models exhibit 33-79% hallucination rates on factual queries, meaning the tools interpreting signals are themselves unreliable on verification tasks. Signal pollution and model hallucination create a compounding authentication challenge no current vendor has solved. Organisational readiness remains poor: 42% of companies have abandoned the majority of their AI initiatives, governance-execution gaps persist, and early 2026 analysis shows organizations consolidating to fewer vendors and shifting from experimental budgets to outcome-based pricing models—signaling tighter scrutiny of ROI and implementation costs. McKinsey's April 2026 survey of 10,000 leaders found 72% of organizations unprepared for upcoming organizational changes, yet horizon scanning capability maturity suggests the issue is not signal detection but organizational capacity to act on signals. Governance has reached boardroom level (reported by Chief AI Officer in March 2026), transforming horizon scanning from technical capability into organizational readiness question. Until signal authenticity can be verified at scale and organizational governance frameworks evolve to operationalize weak signals, horizontal expansion stays blocked.

TIER HISTORY

ResearchJan-2023 → Jan-2023
Bleeding EdgeJan-2023 → Jul-2024
Leading EdgeJul-2024 → present

EVIDENCE (74)

— Center for Strategic and International Studies Risk and Foresight Group conducts continuous horizon scanning across geopolitical, technology, and governance trends; published Global Foresight 2036 survey capturing 450+ geostrategist forecasts on macrotrend evolution.

— Atlantic Council GeoStrategy Initiative operates formalized foresight program with Global Foresight 2036 survey (450+ geostrategists) and snow leopard analysis for underappreciated macro risks; demonstrates sustained institutional adoption of systematic trend identification.

— Silent Eight launched agentic AI system for continuous regulatory and geopolitical horizon scanning, contextually interpreting external developments with transparent reasoning and human-in-the-loop governance; represents production-ready agentic approach to trend monitoring.

— German Bundestag Office of Technology Assessment (TAB) operates systematic horizon scanning since 2014, combining software-based source analysis with expert validation across technological, economic, ecological, social, and political dimensions; sustained institutional deployment.

— Security technology company (G+D) published internal Trendradar methodology tracking weak signals across four innovation domains (data/trust, immersive interaction, intelligent infrastructure, social transformation); demonstrates corporate institutionalization of systematic trend monitoring.

— Empirical HBR-published study testing 7 leading LLMs across 15,000+ strategic decision scenarios; found models converge heavily toward culturally fashionable positions, recommending 'stuck in the middle' strategies rather than contextual analysis—critical negative signal on AI quality in trend-based strategic advice.

— European Union cybersecurity agency (ENISA) published comprehensive methodology for systematic technology signal identification and assessment; represents EU government-level deployment of structured horizon scanning for critical infrastructure preparedness.

— Swiss Academy of Engineering Sciences operates foresight program with federal mandate for early technology identification, producing Technology Outlook platform and situation analyses; demonstrates government-tier deployment of systematic horizon scanning infrastructure.

HISTORY

  • 2023-H1: South Korea's national research institute (KISTI) deployed automated weak signal detection AI for horizon scanning, detecting 439 emerging technologies across 24 science and technology fields. Marked early production use of AI for systematic trend identification.
  • 2024-Q1: Horizon scanning capability expanded beyond research institutions into news aggregation platforms. Halfspace deployed Horizon Scanner for Industriens Fond, demonstrating AI-powered personalized trend discovery from high-volume news sources. IP intelligence vendors (Patsnap, Clarivate) expanded AI search capabilities for patent and literature monitoring.
  • 2024-Q2: Regulatory horizon scanning emerged as distinct vertical application. FinregE deployed RIG for automated regulatory change monitoring with claimed 50-90% productivity gains, indicating vertical-specific commercialization. BERTrend research paper advanced technical methods for weak signal detection using neural topic modeling. General AI adoption headwinds persisted, with adoption barriers in data governance and signal-noise discrimination limiting horizontal expansion.
  • 2024-Q3: Vertical-specific deployments consolidated while broader adoption faced headwinds. IEEE-SA published white paper on AI horizon scanning methodology for standards development, advancing systematic approaches to signal detection. Gartner's August 2024 report warned of 30% GenAI project abandonment risk due to data quality, risk controls, and unclear ROI—highlighting deployment cost barriers ($5-20M for significant transformation). Trend identification remained concentrated in regulated and competitive-intelligence domains.
  • 2024-Q4: IP intelligence and innovation research saw continued commercialization—Patsnap released AI tools delivering 75% faster innovation and 25% lower R&D cost, attracting 18,000+ active users. However, macro adoption headwinds intensified. Appen's survey documented AI ROI decline from 56.7% to 47.3% due to data quality obstacles. Governance gaps widened: Deloitte found 58% GenAI adoption but only 21-59% with controls; financial services at 32% governance compliance. IEEE-SA's updated horizon scanning white paper reinforced methodological rigor but underscored resource intensity. Trend identification adoption remained constrained by data maturity, declining project ROI, and governance control implementation costs.
  • 2025-Q1: Academic and commercial literature on AI-driven foresight expanded. Kingston University published peer-reviewed chapter on causal AI and LLMs for automated horizon scanning; Clarivate released 2025 Research Fronts Report with AI-driven trend detection identifying 128 research fronts from 13,830 candidates. METR benchmark showed AI models doubling task-completion capability every 7 months, reaching 50-60 minute horizon. However, adoption barriers intensified sharply: 42% of organizations abandoned AI initiatives (vs 17% six months prior) due to data quality, leadership misalignment, and $5-20M hidden costs—signaling critical constraints on trend identification deployment despite improving technical capability.
  • 2025-Q2: Vertical-specific tools and methodologies advanced while macro deployment failures accelerated sharply. Research-grade horizon scanning tools (SCANAR, AIDOC) demonstrated 62% reduction in manual review effort in healthcare; regulatory and IP vendors released new offerings (4CRisk product GA, claimed 70% time reduction). EU policy framework adopted horizon scanning for technology foresight (FUTURINNOV project). However, organizational adoption collapsed: 42% of companies scrapped majority of AI initiatives by June 2025 (persistent from Q1), average PoC abandonment reached 46%, and only 4% of organizations reported consistent AI value creation. Technical capability and organizational execution diverged sharply—trend identification tools improved while implementation capacity declined.
  • 2025-Q3: Commercial and government deployments accelerated for vertical-specific and institutional horizon scanning. Clarivate and SAI360 released GA horizon scanning products (5M+ source monitoring); UK Defra deployed horizon scanning capability via Futures Toolkit; FUTURINNOV conducted systematic horizon scanning exercise for EU technology foresight. Research showed 33% of healthcare horizon scanning studies now use automated approaches. Trendtracker demonstrated multi-sector enterprise adoption (Ageas, PepsiCo, P&G, PwC, Siemens, BNP Paribas Fortis, Roularta). Organizational barriers remained unchanged: 42% abandonment rate stable, governance-execution gap widened, horizontal adoption for corporate foresight stalled despite mature tooling. Bifurcation solidified: vertical-specific and institutional applications showed momentum; general business foresight adoption remained constrained by governance and implementation cost barriers.
  • 2025-Q4: Vendor methodologies matured with ITONICS publishing six concrete AI advances for automated weak signal detection and trajectory analysis. OSINT+AI frameworks demonstrated maturity in critical infrastructure security applications. However, a significant new failure mode emerged: signal stream pollution by AI-generated synthetic content (deepfakes, fabricated messaging, bot-amplified signals) now breaks traditional weak-signal detection models and adds signal-authenticity verification as a new adoption barrier. Organizational abandonment persisted at 42%; governance-execution gap remained unchanged. Vertical-specific deployments continued momentum while synthetic signal contamination exposed architectural vulnerability in horizontal business foresight adoption.
  • 2026-Jan: IP and research verticals showed continued maturation: Clarivate's 85% AI adoption in IP ecosystem (up from 57% in 2023) and Nexus product GA addressing trust gaps in research data. FUTURINNOV's ocean observation horizon scanning exercise demonstrated ongoing EU policy application of systematic trend identification. Patsnap's production ecosystem (18,000+ users) sustained metrics of 75% innovation acceleration and 25% R&D cost reduction. However, reliability emerged as critical blocker: latest AI reasoning models exhibited 33-79% hallucination rates on factual queries, directly threatening authenticity verification requirements in horizon scanning pipelines. Professional services adoption visible (Clifford Chance, Slaughter and May) but constrained by signal verification challenges. Vendor product maturity advanced while signal-authenticity crisis widened.
  • 2026-Feb: Horizon scanning systems remained operationally active with multiple production deployments. Clifford Chance continued institutional deployment of structured horizon scanning for regulatory monitoring, tracking policy shifts in Asia-Pacific, Europe, and digital frameworks (Taiwan AI legislation, Vietnam data laws, EU Digital Networks Act). Automated curation systems (Horizon Summary, Champaign Magazine's AI-driven weekly digest) demonstrated real-time trend identification in production. However, a critical information environment degradation emerged: analysis identified 40% of web content now AI-generated, fundamentally compromising weak-signal detection and authenticity verification—the core reliability requirement for horizon scanning systems. This signal contamination combined with 33-79% model hallucination rates created a compounding authentication challenge. Gartner forecasted 40%+ cancellation of Agentic AI projects by 2027, underlining deployment constraints despite operational system maturity.
  • 2026-Q1 (Mar): New vertical deployments and product releases affirmed practice maturity in specialized domains while macro adoption barriers persisted. CPG brand achieved $70M incremental revenue through AI-powered trend identification 4-6 months ahead of competitors (i-Genie). Regulatory vertical expanded via LEGALFLY with 50%+ contract review time reduction across 10+ jurisdictions. Patsnap released Pulse feature for continuous competitor trend and research monitoring (March 2026), signaling vendor momentum. Enterprise platforms (TSC.ai) demonstrated scale across 104+ countries. However, organizational readiness barriers intensified: early 2026 data showed enterprise budget consolidation, shift from experimentation to outcome-based pricing, and persistent work-waste barriers impeding broader adoption. Horizon scanning remained bifurcated: vertical applications thrived while horizontal corporate foresight adoption stayed blocked by signal verification challenges and organizational execution gaps — with roughly 40% of web content now AI-generated, the authenticity verification problem that underlies horizontal adoption continues to compound.
  • 2026-May: Institutional horizon scanning expanded with major think tanks (CSIS, Atlantic Council) and government bodies (Swiss SATW, German TAB, EU ENISA) demonstrating sustained systematic deployment, while Silent Eight launched a production agentic Horizon Scanning Agent with human-in-the-loop governance for continuous regulatory and geopolitical monitoring. Contrasting negative signal: an HBR-published empirical study testing 7 LLMs across 15,000+ strategic scenarios found models converge toward culturally fashionable "trendslop" rather than contextual analysis — confirming that AI-augmented horizon scanning requires safeguards against consensus-bias amplification, not just signal aggregation.
  • 2026-Apr: Institutional-scale deployments and methodological standardization defined the month. UK Defence Science and Technology Laboratory's AI-powered horizon scanning deployment — processing 300K+ articles monthly with analyst hit rates improving from 1% to 40% — won the 2025 Analysis in Government Award, demonstrating validated government-scale deployment. The European Medicines Agency embedded horizon scanning as a strategic regulatory capability; JMIR published the first standardised 35-item reporting checklist for horizon scanning studies, signalling field-wide methodological maturation. Thoughtworks' Technology Radar identified a macro trend shift toward "harness engineering" — infrastructure and feedback loops for AI reliability. Horizon Scan AI platform expanded enterprise reach across UNDP, UNDRR, OECD, and the European Commission. Analyst signals reinforced the ROI correction story: Forrester predicted 25% of planned AI spend deferred to 2027 with only one-third of leaders tying AI value to financial outcomes; Stanford AI Index 2026 documented dual signals (capability breakthroughs alongside 45% spike in AI misinformation and declining public trust). McKinsey survey of 10,000 leaders across 16 countries found 72% unprepared for upcoming structural disruptions, confirming the gap is organizational capacity to act on signals rather than detection capability. IP intelligence tooling matured further: IPRally launched Graph AI patent classification; LexisNexis PatentSight+ extended analytics to 100+ patent-owning organizations; KHネオケム deployed PatSnap Eureka for multi-dimensional business trend analysis. Vertical deployments continue to advance; horizontal adoption remains structurally constrained by signal authenticity challenges and organizational readiness gaps.