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 for cross-functional workflow automation, document processing, and business process optimisation. Evenly split between good-practice and leading-edge: RPA and document extraction are mature; intelligent process mining and autonomous workflow orchestration are still proving out. One practice remains at research stage. Momentum is low — most practices are stalled, with gains coming from incremental automation rather than architectural shifts.
The headline: The automation vendors are profitable. Most of their customers are not. New data confirms only 12% of agentic AI deployments (software that acts on its own without being prompted) deliver strong returns — and the difference is organizational discipline, not which AI model you picked.
Most large enterprises have deployed AI agents somewhere in their operations — 97% by one survey, 83% by another. But having agents running and getting value from them are entirely different things. Independent research across 287 verified deployments shows 540% average ROI for the minority that reach production scale, with payback in under eight months. The problem is reaching production: 78% of companies have pilots, but only 14% have scaled them organization-wide. The automation platform vendors (UiPath, Automation Anywhere, Salesforce) are posting record revenue and profits. Their customers, in aggregate, report single-digit returns. That gap defines where the market sits today.
Gartner forecasts 40% of agentic AI projects will be cancelled by 2027. Based on a poll of 3,400+ organizations, the failures are organizational — wrong use cases, hype-driven scoping, absent governance — not technical. If your AI pilots lack clear ownership, measurable success criteria, and governance frameworks, they are statistically likely to join the cancellation wave.
Forrester quantified the ROI for the deployments that actually work: 540% average, 7.3-month payback. The top quartile exceeded 800%. Customer service leads at 620% ROI. The catch: only 12% of initiatives reach production. The evidence now clearly shows that automation ROI is available but not accessible to most organizations without significant process and governance work first.
Mistral, Orkes, and CamundaCon confirmed the orchestration platform market is consolidating fast. Mistral launched Workflows on Temporal's execution engine (already processing millions of daily runs for ASML, CMA-CGM, Mars Petcare). Orkes raised $60M with tripled customer base. Seven vendors now hold "market leader" status for production-grade agentic orchestration. The window for platform selection is narrowing — 62% of IT leaders already worry about lock-in, and two-thirds of platform migrations fail.
Walmart demonstrated agentic procurement at genuine scale. Using Pactum AI, Walmart's autonomous negotiation agent achieved 3% average savings and 35-day payment term extensions across 2,000+ suppliers with 68-72% agreement rates and 83% supplier satisfaction. This is one of the few documented cases of agentic AI operating autonomously in complex multi-party workflows at Fortune 10 scale.
The product-market fit window for agentic AI vendors is closing. The three major orchestration platforms have all shipped GA products. Databricks telemetry shows multi-agent workflows surged 327% in late 2025. Vendors without two years of production data and an installed customer base face a sharply narrowing path to scale. For buyers, this means commit to a platform architecture before scaling beyond pilots — switching later will be expensive and likely to fail.
EU AI Act enforcement begins August 2026 for worker-facing systems. Any AI managing shift scheduling, performance monitoring, quality control with disciplinary consequences, or resource allocation will likely require governance assessments, oversight documentation, and bias testing. Start your compliance assessment now if you use AI in any worker-management capacity.
"Agent washing" will become visible as cancellation rates climb. Only approximately 130 vendors out of 2,000+ claiming agentic capabilities offer genuine autonomy. As Gartner's cancellation forecast materializes, buyers will demand production case studies, reference customers, and published ROI data. Build your vendor evaluation criteria around deployed evidence, not product demos.
The success rate is low, and it is organizational. 78% have pilots; 14% have scaled. The failures trace to governance gaps (75% lack adequate frameworks), data quality (85% of project failures), and integration debt (48% cite legacy system connectivity as the barrier). Better AI models do not fix these problems.
Orchestration infrastructure matters more than model choice. Analysis of 340 deployments shows the orchestration layer (coordination, state management, audit trails) predicts success 3.2x better than which AI model powers the agents. Most investment flows to model evaluation and prompt engineering — the less important variable.
Silent agentic drift creates a new class of operational risk. An agent that is 95% accurate creates 5% damage at scale, compounding invisibly across thousands of executions. Most organizations lack the monitoring and observability infrastructure to detect drift before it causes material harm, and vendors have not finished shipping the tooling to address it.
Go deeper: the full Operations & Process Automation briefing — the longer analytical write-up, plus every practice we track in this domain with its maturity rating, the tools to consider, and the evidence behind our assessment.