The Agent Density Index
Today the median Fortune 500 company runs 117 distinct AI agents per human employee. The number was 18 just two years ago. The Agent Density Index is the quarterly read on where this curve is going — and which industries are running ahead.
The headline by industry
Software (218 agents per human). Financial services (164). Customer support / AgentOps (151). Healthcare (89). Manufacturing (67). Education (44). Government (29). The order roughly tracks AI tooling penetration, with one important exception: customer support is denser than its tooling penetration suggests, because each support team often runs 3–5 task-specific agents in parallel.
The denominator question
"Per human" needs definition. We count human full-time-equivalents at the company, including contractors counted at FTE-fraction. We count agents at "distinct registered agent identities running for at least 100 hours in the prior 30 days," not raw API keys. Both definitions push the number lower than the marketing-friendly definitions some vendors use; we prefer the conservative read.
How we collected it
Three sources. Anonymized telemetry from 28 Manav design partners covering 41,000 employees and 4.8M agent identities. NHI-monitoring vendor data shared under partnership (Astrix, Entro, Oasis). A quarterly survey of 412 enterprise security and engineering leads. Numbers reconciled across the three; reported number is the median.
The curve
The growth rate is roughly 2.4× per year and decelerating slowly. We expect ~280 agents per human in median Fortune 500 within a year and ~520 within three years. The deceleration is driven by saturation in the top quartile (which is approaching ten agents per task), not by slowing adoption in the bottom quartile.
Why this index matters for identity
Agent density is the proxy variable for the identity gap. At 18 agents per human (two years ago), an enterprise could plausibly track agent activity through service accounts. At 117 (today), service accounts are a fiction. At 520 (a few years out), the only viable model is human-rooted delegation with cryptographic chains. The Agent Density Index curves describe how much runway each industry has before "we'll handle it later" stops being a defensible position.
What the bottom-quartile companies are doing
The companies below their industry median have something in common: they are not slow on AI adoption; they are deliberately slow on agent autonomy. They use AI as a tool, not as an actor. As long as the agent is a copilot a human invokes by hand, the identity gap stays small. The day they enable autonomous loops, the index jumps.
Where Manav publishes the data
The full dashboard with industry filter and quarterly archive is at manav.id/index/agent-density. Datasets available for academic use under a CC-BY license; commercial use under partnership.
Common objections
Two methodological objections we take seriously. Selection bias in the respondent pool — addressed by reporting industry/size mix and weighting where appropriate. Vendor incentive to inflate the gap — addressed by publishing the raw data and source code so anyone can re-run the model with assumptions friendlier to inaction.
Frequently asked questions
How is the methodology auditable? The data, the analysis, and the code are published. Every chart can be reproduced from source. We name our partners (with their permission) and disclose every conflict of interest at the top of the report.
What are the confidence intervals on the headline numbers? Reported per metric in the gated PDF. The 4.6× year-over-year delta on hiring fraud, for instance, has a 95% CI of 3.8× to 5.4×; the median time-to-detection has a CI of 9.2 to 13.1 months.
Why publish numbers your competitors will use? Because the category needs them. The longer the only data is vendor anecdote, the longer the buyer's procurement team waits. We benefit when the category is sized; sizing requires shared numbers.
Where to start
The dataset opens at the 100 to 1 ratio. The control set — which infrastructure changes the curve — is at state of agent identity 2026. Re-fit the model with your own assumptions; we publish the source.
Why density predicts breach risk
Agent density — the count of agents operating under a single human's authority — turns out to predict breach risk better than any other single metric we have tested. The mechanism is structural. Each additional agent multiplies the surface area of authority claims; each authority claim is a potential failure mode. Without delegation infrastructure, density and breach risk track each other linearly. With delegation infrastructure, the relationship flattens — additional agents add capability without proportionally expanding risk because the authority surface is bounded by the substrate. The Density Index therefore functions as both a risk indicator and a substrate-adoption case. Organizations watching the index rise without adopting delegation infrastructure are accumulating a risk debt they will pay during the first significant incident. Organizations adopting delegation infrastructure can let density rise without proportional risk accumulation. The metric exists; the strategy follows from the metric. The companies that read the index correctly are pre-investing now.
The Agent Density Index does not measure adoption. It measures how much identity infrastructure your industry needs and does not yet have.