Accountable for AI you didn't build? We'll prove it's safe.
HumanLens assesses the AI systems health networks, Medicare contractors and banks are held accountable for, vendor-built or in-house, and delivers the signed evidence your regulator, funder or board will ask for.
AI Career Pathway Engine
Frontline workforce navigation · Moses/Weitzman Health System
Start with a single assessment
One AI system, assessed and signed off
Pick the system your regulator, funder or board is most likely to ask about. We run intake, accuracy and bias testing, expert review and sign-off, and hand you a signed system card with the evidence behind it. Fixed scope, fixed price, quoted on the scoping call.
Ongoing governance
Once the first system is signed off, we keep it and the rest of your portfolio on a review cadence, re-test when something changes, and supply the reviews and evidence while your own experts sign. Priced per system covered.
Not sure which system to start with? Score your governance in three minutes →
Three kinds of organization, one shape of problem
The accountability for AI landed on you. The engineering, the vendor's internals and the evidence didn't.
Community health networks
HCCNs and their member centers, answerable under HRSA and Section 1557 for the AI inside their EHRs, ambient scribes and workforce tools. Anchored by our work with Moses/Weitzman Health System.
Medicare contractors and health plans
Organizations accountable to CMS for AI-driven decisions in tools they license rather than build, from claims processing to prior authorization, and expected to show their work.
Mid-size banks
Banks that run AI governance alongside model risk management and answer to examiners for third-party credit, fraud and servicing models they can't open up.
Every assessment, the same five steps
1Governed intake
We inventory the system, who it affects and how decisions flow, then tier its risk on a small set of dimensions your reviewers can defend.
2Accuracy and bias testing
Classical models get accuracy and bias testing against your data. Generative and RAG systems get a scored evaluation set plus a manual expert read of a subset.
3Expert review
Reviewers read what the tier and the test results say they should: the hard calls, the failed checks, the edge cases. Findings come with the evidence attached.
4System card and sign-off
One document your auditor, your board and your engineers can all read. Signed by our expert for a single assessment, by your experts in an ongoing engagement.
5Ongoing watch
Ongoing engagements keep every covered system on a review cadence and re-run the tests when something changes.
Built alongside health systems, newsrooms and funders
"I wish we could go into these reviews knowing that you did all the right things."
Governance lead, Fortune 10 insurer
"Nobody has built a platform for the actual review. Tools exist for pieces of it, but when it is time to get a model approved, the practitioner ends up back in spreadsheets and email."
"I submitted it in January and my next meeting with legal is in March."
From twenty interviews with AI governance executives across healthcare, insurance, finance, media and automotive, attributed by role.
The rules arrived before the review capacity did
EU AI Act obligations, Colorado's AI Act and Texas TRAIGA all expect documented, defensible oversight of AI-driven decisions. Most governance teams have the policies. What they lack is the hours and evidence to prove each system meets them.
Run by people who did this inside a Fortune 5
Before HumanLens, our founders ran responsible-AI review for one of the largest health enterprises in the country and built the data systems behind large-scale learning analytics. That is who reads your evidence.
AI Governance Readiness Check
Eight questions, one score, four dimensions: visibility, speed, defensibility and accountability. See where your program stands before we talk.
Scope your first assessment
Thirty minutes with Cesar to pick the system, agree the scope and set a start date.