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
Autonomous mobile robots, automated storage systems, and goods-to-person solutions that navigate warehouses and fulfil orders. Includes SLAM-based navigation and intelligent order sequencing; distinct from parcel handling which manages shipping and receiving rather than internal warehouse operations. Scope covers AI-driven navigation (SLAM, ML path-planning) and intelligent order sequencing; fixed-path AGVs and simple grid-following robots are out of scope.
Goods-to-person warehouse robotics has matured from a proven practice to an embedded enterprise capability. Systems combining SLAM-based navigation, AI-driven path planning, and orchestrated fleet coordination deliver documented 20-50% service improvements and 25-50% cost reductions at scale, with payback periods between 2.5 and 5 years typical and as short as 24 months for high-frequency tasks. AutoStore, Locus, GreyOrange, and OPEX operate across thousands of sites globally; Locus alone reports 17,000 robots across 360+ sites processing 7 billion picks. The practice boundary has shifted: Fortune 500 operators and increasingly mid-market logistics providers deploy with measurable confidence (DHL-Locus partnership achieved 500M picks with documented 30-180% throughput gains; three independent case studies report 40-65% operational improvements). Yet the defining constraint remains structural, not technological. Mobile manipulation — the frontier of variable-SKU autonomous picking — has reached production deployment through AI-driven gripper technology (Locus NeuraGrasp, Figure AI integration), but adoption execution remains the bottleneck: WES integration complexity adds 20-40% to hardware costs; fleet orchestration requires 300+ robot capacity thresholds to avoid deadlocking; 80% of warehouses remain unautomated despite proven ROI. Organizational readiness, implementation discipline, and vendor integration risk — not hardware capability — now gate broader adoption. Critical limitations persist: lab-to-deployment accuracy degrades from 95% to 60% due to sim-to-real transfer gaps; large-scale deployments still require human exception handlers for multi-SKU kitting and buried items; estimated 60% of supply chain automation initiatives are projected to fail by 2028 due to misalignment rather than technical shortfall.
The market sits at $3.4B in 2026, with goods-to-person systems claiming 47% share and analysts projecting 19.5% CAGR through 2035. Deployment scale has reached inflection: DHL and Locus Robotics achieved a 1 billion pick milestone (March 2026) across 40+ facilities with documented 30-180% throughput gains and 80% reduction in operator training time; Quality Group's European fulfillment center achieved 500,000 picks/day via 350-robot Locus RaaS deployment in <90 days (May 2026); Ocado processes 5 million picks weekly via grid robots and new Porter AMR for pallet handling (MODEX 2026), demonstrating mature multi-modality orchestration; SAP deployed autonomous AI robots in a live Berlin warehouse (May 2026) marking transition from pilots to production integration. Three independent case studies (Maersk, Wooster Brush, brake parts distributor) report 40-65% operational improvements with 6-8 week deployment timelines. Vendor consolidation accelerates capability maturity: Locus Robotics acquired Nexera Robotics (May 2026) to integrate NeuraGrasp, an AI-driven adaptive gripper for variable SKU handling, addressing the long-standing piece-picking bottleneck. Locus Array now operates 17,000 robots across 360+ sites globally, processing 7 billion picks and validating mobile manipulation at enterprise scale. Amazon advances warehouse robotics via next-generation Proteus with conversational interface (25 US centers, European rollout H1 2027), STARK collaborative tote-robot (15 European sites by 2027), and Vulcan tactile sensing; €10B European investment signals enterprise-scale commitment. Humanoid warehousing enters production credibility: Figure AI's Figure 03 completed 200 hours of continuous autonomous operation processing 249,560 packages with zero hardware failures, signaling humanoid fulfillment as emerging parallel to traditional AMRs. The vendor landscape stabilised after 2025 turbulence (Attabotics bankruptcy, Zebra closure), yet order growth in early 2026 significantly exceeded expectations, driven by major retailers (Amazon, Tesco, Marks & Spencer, Decathlon, Apotea). RaaS adoption (64% of companies) democratizes access beyond Fortune 500; SME deployment rose from 23% (2022) to 48% (2025).
Software orchestration has become the critical enabler beyond hardware. Multi-vendor orchestration platforms (GreyOrange GreyMatter, LG CNS PhysicalWorks) now manage fleet coordination at scale; LG CNS documented 15% productivity and 18% cost gains on mixed 100-robot fleets. Amazon's ongoing fleet optimization efforts—DeepFleet AI foundation model achieving 10% efficiency gains—illustrate how orchestration, not raw robot count, determines deployment success. Peer-reviewed research (Transportation Science, INFORMS) confirms that flexible 'swarm' human-robot collaboration—where workers dynamically switch among multiple robots—outperforms fixed one-to-one pairings across the majority of warehouse scenarios, validating organizational flexibility as competitive advantage. The defining tension remains organizational and operational, not technological. Amazon's January 2026 cancellation of Blue Jay ceiling robot underscores expensive iteration paths; adoption satisfaction gap persists (44% deployed, 34% satisfied). Industry expert assessments document the credibility gap: sales-driven vendor cultures, misaligned processes amplified by automation, poor implementation discipline, and career risk discouraging standardization drive widespread adoption failures despite technical maturity. Operational reality at scale reveals critical constraints: mechanical failures emerge at >500K cycle thresholds, hardware lifespan limits (4-6 years for GPUs and control boards) create ongoing obsolescence planning, and supervised autonomy (human-in-loop) remains the permanent operating model. Integration costs 20-40% of hardware; only 12% of warehouses run autonomous picking despite automation availability; only 26% projected to have any automation by 2027. Fleet-level constraints are concrete: Amazon caps utilization at 65% to prevent deadlock; GreyOrange identifies 300-robot ceilings for traditional management; operator-level evidence (1,200+ workers across 5 US facilities, MODEX 2026 survey) confirms robots dominate high-repetition tasks while humans handle multi-SKU kitting and exception handling. Critical barriers persist: WES integration complexity, customer selection of wrong architectural models, insufficient process standardization before deployment, organizational misalignment, and vendor-lock-in risk combine to cause estimated 60% of supply chain automation initiatives to fail by 2028. The technology is proven; adoption barriers are structural, organizational, and economic.
— Amazon advancing warehouse robotics via Proteus conversational interface, STARK collaborative tote-robot (15 European sites by 2027), Vulcan tactile sensing; €10B European investment confirms enterprise-scale commitment.
— Transportation Science (INFORMS) study of warehouse picking configurations: swarm policy (workers dynamically switch robots) delivered higher throughput than fixed one-to-one assignments in majority of operational conditions.
— Ocado MODEX 2026 case study: 5M weekly picks via overhead grid + mobile fulfillment units (Chuck, Porter); new Porter AMR solves pallet handling bottleneck; 30-sec battery swaps reduce downtime, demonstrating mature multi-modality integration.
— Operational reality of deployed fleets (2B+ lifetime picks): mechanical failures unreplicable in labs, GPU/board lifespan constraints (4-6 years), supervised autonomy as permanent model, customer support as critical as hardware.
— Humanoid warehouse milestone: 200 hours autonomous operation, 249K packages, zero failures; robots autonomously self-routed to charging stations, signaling credibility inflection for humanoid fulfillment automation.
— Industry expert roundtable documenting adoption failure patterns: under-qualified sales, VC-driven pressure, automation amplifying broken processes, career risk discouraging standardization. Critical signal on organizational barriers vs. technical readiness.
— Locus RaaS deployment at fast-growing sports nutrition retailer: 500K daily picks via 350-robot fleet scaled from 40 robots in <90 days, validating mobile orchestration at scale with variable-SKU complexity.
— Locus Robotics acquired Nexera to integrate AI-driven adaptive gripper (NeuraGrasp) for variable SKU handling, consolidating capability maturity in mobile manipulation at warehouse scale.