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 supporting, retaining, and understanding customers after the sale. The highest concentration of good-practice tiers: chatbots, ticket routing, sentiment analysis, and voice-of-customer are deployed at scale in most industries. Bleeding-edge frontiers include autonomous resolution without human escalation and real-time emotion detection. Momentum is steady but churn prediction and proactive outreach remain stalled.
The headline: AI for customer service is production-ready and the unit economics now work — but only one in ten companies has actually got it running properly, and consumers don't want it.
AI tooling for customer operations is ready to ship across every major vendor and almost every use case — agent assist, chatbots, call summaries, quality monitoring. Most companies have invested. Few have actually operationalized it. The defining ratio: 82% of customer experience leaders report AI investment, but only 10% have reached mature deployment. The leaders (Klarna, Salesforce, Bank of America) are resolving 70–84% of customer queries without a human and have documented the return. The median company is still piloting. If you've deployed AI agent assist or routing at scale with proper oversight, you're ahead of roughly 90% of peers. If you're still piloting, you're with the majority — but the gap to the leaders is widening.
The unit economics of fully-automated customer service are now provable. A leading customer-service AI platform documented resolving 84% of cases without a human across 380,000+ interactions, at $0.62 per resolution versus $7.40 for a human agent. But only 12% of deployments clear a 300% return — what determines which side of that line you land on is deployment discipline, not which vendor you picked.
Voice AI's gap between demo and reality got sharper. Intuit rolled out Amazon's voice AI across 11 countries handling 275M+ calls a year, proving it works at enterprise scale. But analysis of 10M+ live calls showed response time more than doubling between pilot and production, with customer satisfaction (CSAT) drops and a tripling of escalations to humans. The demo-to-production gap is wide.
Email got autonomous AI agents. Zendesk released — out of beta and generally available (GA) — AI agents (software that acts on its own, here taking multi-step actions like resolving a ticket end-to-end) that handle email, not just chat. This widens what AI can take over, and equally widens what you have to govern.
Insurance claims went from mostly-human to mostly-machine. Across seven major US carriers, the share of claims processed end-to-end without a human jumped from 10–15% up to 70–90%, with resolution 75% faster and 30–40% cheaper. The use case is no longer experimental.
AI is starving the systems that measure customer feedback. Forrester analysts warned that AI agents are absorbing the very interactions that generate Voice of Customer (VoC) data — the survey responses and call transcripts that tell you how customers feel. Survey response rates already dropped from 30% to 18% in six months. If you rely on customer feedback to set strategy, the input pipe is shrinking.
EU rules are quietly reshaping how you can deploy AI in Europe. AI workflows (where humans stay in control) outnumber autonomous AI 5-to-1 in Europe, and 78% of organizations cite compliance as the main barrier to going fully autonomous. The UK's competition regulator (the CMA) can already fine up to 10% of global turnover. If you operate in Europe, get a compliance audit on your autonomous-AI roadmap now.
Half the companies that fired humans for AI may be hiring them back by 2027. A 2026 industry survey predicts 50% of firms that cut service staff to deploy AI will reverse course within a year. Klarna already did, after customer satisfaction collapsed. Build reversibility into any AI-driven headcount plan — the track record on permanent cuts is poor.
Your knowledge base is the lever, not your AI vendor. New peer-reviewed research showed that improving the quality of the underlying knowledge base lifted AI accuracy by 17–23 percentage points — a bigger gain than switching to the most capable AI models (the "frontier models"). If you're spending on chatbots or agent-assist tools without equivalent spend on cleaning up the knowledge base, you're optimizing the wrong layer.
The 10% maturity wall. Investment is at 80%+, mature deployment is at 10%. The blocker is governance, change management, and data quality — not technology. One major vendor's AI revenue hit $500M while only a quarter of its customers had fully integrated the AI.
Augmentation works; full autonomy sells. Designs where a person reviews each AI output before it ships ("human-in-the-loop") consistently beat fully autonomous ones on reliability and customer satisfaction. But vendors price by outcomes and per-resolution, which pushes you to take the human out. 90% of leaders are uncomfortable letting AI represent their brand directly anyway.
Consumer trust is the ceiling. 64% of consumers prefer companies not use AI for service, and AI customer service fails at four times the rate of other AI applications. Faster deployment without quality improvement accelerates brand damage, not cost reduction.
Go deeper: the full Customer Operations 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.