AAI snapshot · Overall readiness
2026-W45 · scale 0–100
Awareness
0–20Active
21–40Operational
41–60Systemic
61–80Transformational
81–100
Organization
72
Brand
45
Product
68
Composite = unweighted average of its five dimensions. Overall = unweighted average of the three composites. Bands computed at render.
Evidence this issue · 24 cited sources across 36 signals · 9 primary-source (38%) · 15/15 dimensions carry a cited signal · 6 dimensions moved
¶
This week
What moved the register
This was the week the enterprise AI number got big enough to argue about. Menlo Ventures put deployment spend at $176.8B by September, up 5.4x from $33B, and the single most revealing line in the report is that agents now generate 80% of all AI tokens, roughly seven times human volume. The consumer of enterprise compute is now the agent. That reframes every dimension in this index. The live question is whether the firm can stand behind what its agents do when nobody is reading every line.
The spend is real and so is the attrition underneath it. Gartner's projection that more than 40% of agentic AI projects will be canceled by the end of 2027 sits directly against Google Cloud's finding that 94% of agent deployers report measurable cost and revenue contribution. Both are true. They describe two populations. One has governed data and an agent inventory, the other has impressive demos that break on the long tail. AvePoint names the mechanism precisely: a program with no agent inventory cannot connect cost to outcome, so it cannot prove value and cannot survive a CFO review. The fault line is governed data and traceable identity. That is why Scaling Maturity, ROI Impact, Agentic Deployment and Value Delivery all held flat this week. The positive and the setback evidence were both strong, and an honest index holds the line through a real divide.
Where the index did move, the Identity Control Surface explains the move. Snowflake's Cortex AI Gateway now authenticates agent tool calls with the end user's identity and checks them against policy before execution. The same week, the UK's DRCF told enterprises plainly that an agent's autonomy does not diminish the deployer's legal responsibility. Governance tooling and governance enforcement arrived together, which is what a durable gain looks like. Apple's annual filing, warning that agent errors may be permanent and difficult to reverse, states the risk that Microsoft Research's PDD Kit answers by proving agent-written code meets developer intent for every input. The Compiled Corporation only works when the compiled decision can be audited back to an accountable identity.
The Decision Surfaces lens reads the infrastructure consolidation the same way. Google folded all Vertex AI services into a single Gemini Enterprise Agent Platform, standardizing where the human-agent interface sits. The platform war is becoming a governance war.
Two cautions close the week. API spend is concentrating hard, Anthropic to 55%, Google down to 9%, which rewards perceived competence and will punish AI-washing. And the public is pushing back, with voters approving restrictions in 40 of 41 local data center ballot measures. The spend curve and the trust curve are diverging. Firms that engineer identity governance into their agents now will be the ones still deploying when the enforcement arrives.
¶
Score movement · week over week
What moved this week
6 dimensions shifted. The trend is the product.
▲ Organization · Workforce AI Access
Workforce AI Access
+1 · 73 → 74
Company AI use climbed to 65% from 55% in the Menlo Ventures survey, with per-company spend driving the inflection and Microsoft enabling Claude by default across Copilot surfaces.
▲ Organization · Governance & Ethics
Governance & Ethics
+1 · 78 → 79
Snowflake's Cortex AI Gateway now authenticates agent tool calls with end-user identity and checks policy before execution, while UK regulators state the deployer stays legally accountable for agent actions. Tooling and enforcement moving together.
▲ Brand · Agent-Ready Infrastructure
Agent-Ready Infrastructure
+1 · 63 → 64
Google Cloud consolidated all Vertex AI services into the Gemini Enterprise Agent Platform, and Lumen and Microsoft expanded co-selling to connect data, apps and network for AI workloads.
▲ Brand · Content AI-Readiness
Content AI-Readiness
+1 · 37 → 38
Investment shifted to semantic layers and governed context after analytics agents were found to fail mainly on lack of business context, restructuring content for machine interpretation.
▲ Product · AI Interaction Layer
AI Interaction Layer
+1 · 76 → 77
Agents now generate 80% of all AI tokens, roughly 7x human volume, with new agent apps from OpenAI, Meta and Anthropic shipping in the window. The interaction model moved decisively up the agentic spectrum.
▲ Product · AI UX Maturity
AI UX Maturity
+1 · 60 → 61
Microsoft Research's PDD Kit formally proves agent-written code meets developer intent for every input, a direct explainability gain against Apple's stated concern that agent errors may be irreversible.
¶
The full read · 15 dimensions
Fifteen dimensions, three categories
Cell shade tracks the score; the band label carries it in words, so nothing rides on color alone.
Organization72 · Systemic
Lower
Higher · shade within each category hue
¶
Organization · Workforce AI Access
Workforce AI Access
Systemic · 74/100 · +1 week over week
Workforce access widens up a point. The Menlo Ventures survey of roughly 400 US enterprise decision-makers puts the share of companies using AI at 65%, up from 55%, with per-company spend driving the inflection as teams hand agents longer-running work. Microsoft's enabling Claude models by default across Copilot, Researcher and Copilot Studio in commercial cloud widens the set of models workers touch daily. The move is structural, workers are reaching for assistants inside existing workflows.
¶
Organization · Scaling Maturity
Scaling Maturity
Systemic · 70/100 · no change week over week
Scaling holds flat. The week's evidence pulls both ways and nets to no move. Positive: the Google Cloud ROI study finds mature agentic deployments accelerating returns. Negative: more than 40% of agentic AI projects are projected for cancellation by the end of 2027 on escalating cost and unclear value, and AvePoint reports 86.9% of organizations delayed generative AI deployments over data-quality issues. The fault line is governed data. Firms with clean data foundations scale, the rest run impressive demos that never reach production. A flat week is the honest read.
¶
Organization · Governance & Ethics
Governance & Ethics
Systemic · 79/100 · +1 week over week
Governance advances a point. The Identity Control Surface is where this week's signals cluster. Snowflake's Cortex AI Gateway now authenticates agent tool calls with the end user's identity and checks them against policy before execution. UK regulators through the DRCF state plainly that an agent's autonomy does not diminish the deploying organization's legal responsibility, forcing non-human identity governance onto the board agenda. The tooling and the enforcement signal are moving together, which is what a durable governance gain looks like.
¶
Organization · Talent & Upskilling
Talent & Upskilling
Systemic · 72/100 · no change week over week
Talent holds flat. The week supplies demand-side evidence without a measurable shift in enterprise fluency. MIT research cited by Forbes names the last-mile engineering gap as the reason most pilots produce no P&L impact. Google.org's funding of AI training for teachers shows upskilling investment at the education layer. The signal confirms where the skill shortage sits. The workforce has not yet closed it, so the score holds.
¶
Organization · ROI Impact
ROI Impact
Systemic · 66/100 · no change week over week
ROI holds flat. Two strong, opposing signals cancel. Google Cloud's survey of 2,400+ executives reports 94% of agent deployers seeing measurable cost and revenue contribution and 84% reporting increasing returns. Against that, MIT finds most enterprise pilots produce no measurable P&L impact, and AvePoint reports the CFO has moved from budget owner to value gatekeeper as untraceable agent spend compounds. The return is real where deployment is mature. The distribution has not shifted enough to move the score.
¶
Brand · AI-Native Messaging
AI-Native Messaging
Operational · 47/100 · no change week over week
Messaging holds flat. Atos moved from Market Challenger to Leader in the ISG Provider Lens for agentic AI services, pairing its claims to governed, production-grade delivery through the Sovereign Agentic Studios. That is a Janus Brands positive, legacy integrator credibility aligned with current agentic positioning. One analyst recognition does not move the category baseline, so the score holds.
¶
Brand · Agent-Ready Infrastructure
Agent-Ready Infrastructure
Systemic · 64/100 · +1 week over week
Agent-ready infrastructure advances a point. Google Cloud folded all Vertex AI services into the Gemini Enterprise Agent Platform, making build, scale, govern and optimize a single machine-readable foundation. Lumen and Microsoft expanded co-selling to connect the data, applications and network behind AI workloads. The Decision Surfaces lens reads this as the agent interface standardizing, which is the precondition for structured, API-addressable enterprise content.
¶
Brand · Market AI Perception
Market AI Perception
Active · 36/100 · no change week over week
Market perception holds flat. The week's evidence is mixed and nets to no move. Enterprise API spend is concentrating sharply, Anthropic to 55%, Google down to 9%, which rewards perceived competence at the top and punishes the field. Against that, public backlash hardened, with voters approving restrictions in 40 of 41 local data center ballot measures and tech-worker organizing rising. Credibility is bifurcating. The category average does not move.
¶
Brand · Content AI-Readiness
Content AI-Readiness
Active · 38/100 · +1 week over week
Content readiness advances a point. The top reason analytics agents produce weak insights is lack of business context, and the response is investment in semantic layers and governed context. TrustHouse.AI's framing of data, business and process context as engineered-together infrastructure points the same way. Content is being restructured so agents can read meaning. Early but directional, which supports a modest move.
¶
Brand · AI-First Orchestration
AI-First Orchestration
Active · 38/100 · no change week over week
Orchestration holds flat. Microsoft's Logic Apps Automation describes adaptive orchestration that combines contextual AI decisions with organizational knowledge and human oversight for processes too variable to fully predefine. The evidence is tangential to brand and marketing operations specifically, so the prior score holds.
¶
Product · AI Interaction Layer
AI Interaction Layer
Systemic · 77/100 · +1 week over week
The interaction layer advances a point. Agents now generate 80% of all AI tokens, roughly 7x human volume, per OpenRouter data in the Menlo report. The interaction model has moved decisively up the chatbot-to-agentic spectrum, with new consumer agent apps from OpenAI, Meta and Anthropic shipping in the window. The human-to-assistant interface is giving way to agent-to-agent consumption, which is the defining shift this dimension tracks.
¶
Product · Agentic AI Deployment
Agentic AI Deployment
Systemic · 77/100 · no change week over week
Agentic deployment holds flat. Real production evidence this week, agent spend inside the $176.8B deployment total and Honeywell's Nemotron agents running at Dangote's Lekki refinery. The offsetting signal is strong: Gartner's projection that more than 40% of agentic projects are canceled by 2027 on a POC-to-production validation gap, and the OpenAI-Hugging Face probe record of operational-safeguard failures over a 64-day sequence. The positive and the setback are both material. The dimension holds.
¶
Product · AI UX Maturity
AI UX Maturity
Systemic · 61/100 · +1 week over week
UX maturity advances a point. Microsoft Research's PDD Kit converts developer intent into a formal specification and proves agent-written code meets it for every input, letting humans review intent instead of every line. That is a direct gain in explainability and trust at the agent boundary. Apple's filing warning that agent errors may be permanent and difficult to reverse names the risk this tooling addresses. Verification infrastructure arriving against a clearly stated reversibility problem supports a modest move.
¶
Product · Process Redesign
Process Redesign
Systemic · 61/100 · no change week over week
Process redesign holds flat. Applied Materials' whole-fab digital twins show genuine workflow rebuild, connecting production data to models that evaluate changes across the factory. The countervailing signal is that most pilots validate only the horizontal layer and break on the long tail of real exceptions. Deep redesign is happening in physical-AI leaders and stalling elsewhere. The two cancel, so the score holds.
¶
Product · AI Value Delivery
AI Value Delivery
Systemic · 66/100 · no change week over week
Value delivery holds flat. Google Cloud's finding that 94% of agent deployers report measurable cost and revenue contribution is strong positive evidence. It is matched by AvePoint's point that governance gaps make value untraceable to a specific agent, so programs without an agent inventory cannot connect spend to outcome. Value is real and measurable where identity and lineage are in place, and unprovable where they are not. The score holds.
For agents · priced data cuts
This issue is free to read in full. Agents that need a specific historical or category cut on demand settle it over x402: USDC on Base mainnet, gasless, the signed data returned in the same exchange. Prices render from live route config.
GET /x402/ai-index/compositesThree composites, with history0.005 USDC
GET /x402/ai-index/organizationOrganization category · 5 dimensions0.01 USDC
GET /x402/ai-index/brandBrand category · 5 dimensions0.01 USDC
GET /x402/ai-index/productProduct category · 5 dimensions0.01 USDC
GET /x402/ai-index/dimensionAny single readiness dimension0.005 USDC
Free, always: this canonical issue, every claim endpoint, the deltas, and the methodology.
Methodology & sources
15-dimension enterprise AI readiness framework scored 0-100 across Organization (5), Brand (5), and Product (5) categories. Scores derived from public signals: news and company announcements found by web search, and the published feeds of AI vendors, research labs, and the technology press. Dimensions equally weighted within each category.
Awareness 0–20Active 21–40Operational 41–60Systemic 61–80Transformational 81–100
Produced by Applied Identities using specialized AI analysis. All scores based on publicly available data. Internal deployment data not captured. This analysis is independent and does not represent the views of any company named herein.
Produced by 3Jane Intelligence under the Applied Identities byline · methodology 1.0. Sources are cited as sources, not endorsements. The full machine payload is at /ai-index/2026-W45/payload.json.