AI you can stand behind.
Know every AI tool your team uses, set the rules, train your people and prove it. Your readiness score, inventory and report are free with any ETA account.
Your team already uses AI. Can you show how it's governed?
Shadow AI is everywhere
People sign up for chat assistants, note takers and image tools with work email. Customer data goes with them. You can't govern what you can't see.
The rules are arriving
The EU AI Act's AI literacy duty and banned uses already apply. Colorado's AI law starts January 1, 2027, and EU high-risk rules follow in December 2027. Customers and funders are asking too.
Governance is mostly people
A clear policy everyone has signed, training everyone has finished and a way to report problems prevent more harm than any tool. Steward makes all three easy to run and easy to prove.
Eight tools, one place
AI inventory
Every AI tool and use, with a risk tier and what that tier asks of you. Start from a catalog of 52 common tools.
Shadow AI finder
Scan your Google Workspace or Microsoft 365 app export to find AI tools nobody approved.
Readiness score
24 questions mapped to NIST AI RMF, ISO/IEC 42001, the EU AI Act and US state laws, with a 90-day plan.
Policy builder
A plain-language AI acceptable use policy drafted from your answers and your approved tools.
Staff sign-off
Each person gets a private link to read the policy and sign with their name. No account needed.
Training tracking
Who has finished AI training, checked against ETA credentials. Start with ETA's free AI Fluency course.
Vendor reviews
Fourteen questions on data use, contracts and security, scored, with a date for the next review.
Guardrails, incidents and a trust page
Check text before you paste it, hold ETA Switchboard agents to your rules, log incidents and publish how you use AI.
A 40-person services firm, 20 minutes in
Four AI tools from the catalog, a few honest answers, and Steward's real output.
Their AI inventory
- ChatGPTLimited risk
- Fireflies.aiLimited risk
- Website chat assistantLimited risk
- Resume screeningHigh risk
What resume screening asks of them
- Do an impact assessment before use and each year after: purpose, data, who it affects, known risks and how you limit them.
- Tell people at the point of use that AI is part of the decision.
- When the outcome goes against someone, explain the main reasons and let them correct their data and ask a person to review it.
- Test for unfair results across groups, and keep records of decisions and testing for at least 3 years.
- NIST AI RMF 1.0
- 32%
- ISO/IEC 42001
- 33%
- EU AI Act
- 46%
- US state laws
- 8%
First 30 days of their plan
- Tell people how to report AI problems within a set time, and log every report.
- Review each AI vendor and set a next review date (6 months for high risk, 12 otherwise).
- Write an impact assessment for each high-risk use: purpose, data, people affected, risks and safeguards.
Areas scored: Govern, Map, Measure, Manage.
Your score is free. The full program comes with your membership.
Steward comes with your ETA account. Organization memberships get a shared workspace for the whole team.
Free ETA account
Free
- Readiness score mapped to four frameworks
- 90-day plan and report (PDF or Word)
- AI inventory with risk tiers, up to 10 tools
- Check-before-you-paste for personal data and secrets
- 2 vendor reviews
Explorer and Student
Included
- Everything free, plus the AI policy builder
- Staff sign-off with personal links
- Training tracking with ETA credential checks
- Incident log and impact assessments
- 100 tools, 10 vendors, 25 people
Pro, organization memberships and sponsored seats
Included
- Everything in Explorer
- Shadow AI finder for Google Workspace and Microsoft 365 exports
- Public AI trust page
- A shared workspace for your organization
- 1,000 tools, 250 vendors, 2,000 people
Sponsor AI governance for the organizations you serve
Economic development organizations, counties and states can give small businesses, schools and agencies Steward with a program code, alongside AI Ready training. Participants opt in to share their readiness score, and you see the combined number for your cohort, never their policies, tools or people.
Works with the rest of ETA
- Plan and fund AI projects with ETA IntelliLaunch.
- Run AI agents with guardrails in ETA Switchboard; Steward checks them against your rules.
- Get your data AI ready with ETA Groundwork.
- Train your team with ETA's AI courses.
Good to know
Does ETA see our inventory, policy or people?
No. Your workspace is private to you or your organization. ETA staff see counts, and help requests you choose to send. A program that sponsors your seat sees only your readiness score, combined with others, and only if you opt in.
Is the policy legally reviewed?
The policy builder drafts a clear, current policy from your answers, and the readiness questions are mapped to the main frameworks and laws (reviewed November 2026). It isn't legal advice: have counsel review your policy and obligations, or ask ETA to connect you with a vetted provider.
What does the shadow AI finder do with our export?
It reads the file in memory, matches app names and web addresses against a catalog of known AI tools, and discards it. Steward keeps only the tools you choose to add.
Does 'check before you paste' send our text anywhere?
No. It runs in your browser. It flags emails, phone numbers, ID and card numbers, keys and hidden instructions and gives you a masked copy.
Do people need an ETA account to sign the policy?
No. Each person gets a private link to read the current version and sign with their name. Steward never emails them; you send the links yourself.
Can our whole organization use one workspace?
Yes. With an ETA organization membership (Startup, Nonprofit, Corporate, Partner or Sponsor), everyone with a seat works in one shared workspace, and your ETA admin publishes the policy and trust page.
ETA Steward helps you organize your AI governance. It isn't legal advice: have counsel review your policy and obligations before you rely on them.
See how ready your AI program is
Free with any ETA account. Answer 24 questions and get your score, framework coverage and a 90-day plan.