The state of agent readiness in 2026
We built evidence files on 4,350 companies across 22 industries and scored what we could verify, not what companies claim. Here’s what we’ve learned.
“AI adoption is easy to announce. Production systems are harder to find. We built this benchmark to score the evidence we can see, not the claims. Learn what separates the ready.”
Findings at a glance
Readiness falls mid-market
Mean readiness by employee band
Scale does not translate into steady readiness gains. Companies with 2,000–10,000 employees are the market’s low point. The same dip appears in nearly every industry.
Biggest gaps are within industries
10th to 90th percentile score range; dot = median
Professional Services spans 61 readiness points from its 10th to 90th percentile. That is wider than the gap between the highest- and lowest-scoring industries.
The stack isn’t the bottleneck
Weakest pillar by company
Nearly every company has modern data infrastructure. Yet talent or production controls are the weakest area for 70% of companies.
Everyone announces AI. Evidence is harder to find.
Most companies have the infrastructure to support AI. Far fewer have the people and controls to run agents in production. And the public record of 3,075 verified announcements over two years is moving faster than the work behind it.
So we scored evidence instead of claims: 304 detectable technologies, the hiring pipeline, workforce composition, and the official press record. No surveys, no self-reporting. The result is a single agent-readiness score, five evidence pillars, and five readiness levels.
The largest public evidence-based analysis of company AI readiness ever published
Six findings worth the download
Talent and controls lag the stack, even at the top
99.7% of companies show a modern-stack signal. Yet among level-5 companies, 91% show an agent-builder platform and only 66% show evaluation tooling. The constraint isn’t technology.
Why readiness falls mid-market, and what the companies are missing
Digital natives score 61–71 at every company size. Incumbents dip hard between 2,000 and 10,000 employees. Two comparisons explain what gets lost on the way up.
The biggest gaps are within industries, not between them
Industry averages vary 2.6×, but Professional Services alone spans 61 points from its 10th to 90th percentile, wider than the gap between the best and worst industries.
The press record: loud talkers, loud builders, quiet builders
Of 40 companies that publicly announced AI governance, 75% show no evaluation evidence. Meanwhile some of the strongest builders show no detected press at all.
Business model and regulation explain less than you think
The B2B/B2C gap disappears after controls. And regulated industries show no more evidence of production controls than anyone else (p=0.66).
The top 50, and the five executive priorities behind them
From JPMorganChase (266,000 employees) to Plaud (141). Scale isn’t a requirement: the ranking rewards balanced evidence, and the agenda that produces it is coachable.
Download today
19 pages of original research. 33 charts. An evidence-only benchmark of 4,350 companies, based on real signals.
- Score distribution across 5 readiness levels and 5 archetypes
- Pillar-level diagnosis: where talent and controls lag the stack
- Say-versus-do: 3,075 announcements classified for substance
- The top-50 leaderboard, from JPMorganChase to Plaud
Three audiences, one shared blind spot
Benchmark your company against 4,350 peers
Find your readiness level and your weakest pillar. Pressure-test your roadmap against the five executive priorities that separate levels. The evidence that matters changes as you climb.
You’re likely in the dip
The 2,000–10,000 employee dip appears in nearly every industry. See what digital natives keep at that size that incumbents lose, and which signals close the gap.
The announce-to-build ratio belongs in due diligence
Separate loud talkers from quiet builders. Only 2.8% of announcements describe production systems. The press record alone won’t tell you who is actually ready.
A sneak peek at the insights
Mean readiness by employee band and company type
Digital natives stay resilient
Digital natives score 61 to 71 on agent readiness at every company size. The mid-market decline comes almost entirely from incumbents.
Claim classification across 3,075 verified events
Agent claims everywhere, evidence is not
61% of verified AI announcements mention agents, but only 6% claim a production system and 3% mention governance.
Mean readiness by company age, incumbents only
Company age does not map to readiness
Among incumbents, companies founded between 1976 and 2001 score highest, while century-old businesses outperform the postwar cohort.