A finance operating model for where AI creates value, where risk forms, who owns the lever, and what evidence proves progress.
Most AI-adoption models are too narrow to act on, or too broad to diagnose. Twelve is where the model stops flattening the picture.
We arrived at twelve backwards from real CIO, CISO, CFO, and CEO conversations. The point is not a scorecard. It is to let an executive look at their organization and know, in under ten minutes, which levers they are pulling well, which they are not, and which they have not thought about. Use the maturity assessment when the question is where to start.
Twelve levers. One operating model.
Each lever names a question the buyer is already asking. Pull them well and AI compounds. Miss one and the operating plan loses traction.
Discovery & inventory
A live, consolidated list of every AI tool, embedded feature, and internal agent in use.
Usage depth & breadth
Session length, feature engagement, workflow integration. Not login counts.
Workflow integration
AI inside the IDE, CRM, spreadsheet, inbox. Not in a side browser tab.
ROI & value attribution
Cost-per-outcome attribution that connects spend to cycle time, throughput, quality.
Training & enablement
Role-fit tooling and prompt patterns the employee uses inside their workflow the same week.
Shadow AI management
Discovery, classification, and policy enforcement on AI that never went through procurement.
Spend intelligence
License utilization, duplicate-tool detection, individual subscriptions on retail plans.
Policy & governance
Acceptable-use, data-handling, model-selection policies as code, applied at the trace.
Agent behavior evaluation
Production-trace evaluation: groundedness, tool-call authorization, drift, baseline tracking.
Cross-org visibility
One pane of glass across vendor tools, embedded AI, and internal agents.
Change management
Role design, manager enablement, OKR alignment. The work that moves pilots into production.
Benchmarking & maturity
Cross-customer baselines so the CIO knows whether the org is ahead, at par, or behind.
Four anchors. Twelve handles.
The AI Audit produces the visibility every other anchor needs. Transformation captures the upside. Governance contains the risk. Fluency builds the workforce capability that compounds the rest.
Pulled together, the twelve handles let an enterprise run the AI work from one operating model rather than five disjoint workstreams.
Which category of tool pulls which levers.
Categories, not company names. Use this to stress-test any vendor that claims full coverage.
LEVER | ADOPTION ANALYTICS | DEV GOVERNANCE | SHADOW AI SECURITY | PLATFORM VENDORS | CONSULTANCIES | TRUSTEVALS |
|---|---|---|---|---|---|---|
1. Discovery & inventory | ◐ | ○ | ● | ○ | ◐ | ● |
2. Usage depth & breadth | ● | ○ | ○ | ◐ | ◐ | ● |
3. Workflow integration | ◐ | ○ | ○ | ◐ | ◐ | ● |
4. ROI & value attribution | ◐ | ○ | ○ | ○ | ◐ | ● |
5. Training & enablement | ◐ | ○ | ○ | ◐ | ● | ● |
6. Shadow AI management | ○ | ○ | ● | ○ | ◐ | ● |
7. Spend intelligence | ● | ○ | ◐ | ○ | ◐ | ● |
8. Policy & governance | ○ | ● | ◐ | ◐ | ◐ | ● |
9. Agent behavior evaluation | ○ | ◐ | ○ | ◐ | ○ | ● |
10. Cross-org visibility | ◐ | ○ | ◐ | ○ | ◐ | ● |
11. Change management | ○ | ○ | ○ | ○ | ● | ● |
12. Benchmarking & maturity | ◐ | ○ | ○ | ○ | ◐ | ● |
● covered · ◐ partial · ○ not covered
Three more from the practical guide.
The Baseline Problem
Frameworks tell you what to track. They don't tell you what 'good enough' looks like.
From Adoption to Assurance
Why governance demand follows adoption proof, not the other way around.
Why Continuous Beats Periodic
What governance looks like when the system being governed is non-deterministic.
Twelve levers. Four anchors. One operating model the CIO can walk into a board meeting with.
Bring one workflow, vendor, or AI portfolio. We will map the evidence needed for finance leaders to fund, ship, or stop it.
Common questions on the 12 Levers.
A four-cluster framework: Visibility, Adoption, Evaluation, and Governance. It names every lever an enterprise has to pull to run AI at scale. Each lever has a question it answers, a definition of 'good,' a named owner, and a category of tool that covers it.
Twelve is where the model stops flattening the picture. Three leaves things off; twenty is too broad to be diagnostic. Twelve was arrived at backwards from real CIO, CISO, CFO, and CEO conversations: the questions that get asked, the gaps that get flagged.
Discovery, Usage, Workflow, ROI, and Spend typically sit with the CIO or CAIO. Shadow AI and Policy sit with the CISO. Training and Change Management sit with HR or a transformation office. Agent Behavior Evaluation is shared between platform engineering and the AI risk owner.
NIST and ISO are management-system frameworks. They tell you what to track. The 12 Levers is an operating model. It tells you what to do, who owns it, and what evidence proves progress. Use both: the framework for compliance, the levers for execution.
Print the twelve levers. Walk into each function: IT, Engineering, Ops, Finance, Compliance. Ask which of these they own, partly own, or do not own. The gaps in that grid are usually the most interesting finding of the quarter.
A 15-page printable edition is linked from the hero CTA. Email gate, one-time. Cover, four-cluster overview, one page per lever, closing CTA. Designed to forward to a CIO, CISO, CFO, or board sponsor.
Related links and sources
- PILLAR The Baseline Problem Frameworks tell you what to track.They don't tell you what 'good enough' looks like. READ THE ESSAY →