Trust and security
AIVROs measures how ready a company is to use AI and where AI will pay. That only works if people answer honestly, so privacy is built into how the product counts, not added on top. This page says what we do, in plain words, and what we have not built yet.
Last updated: October 2026Our promise
- Group results by default. Leaders see figures for groups of five people or more who answered. A smaller group shows as hidden.
- Individual scores are private by default. A workspace owner can choose to see them only after reading and accepting a disclaimer. Every time this is turned on or off, we record who did it and when.
- No tracking software. We read what people choose to answer and what the sponsor uploads. Nothing else.
- We don't train on your data.
AI Safe Use
Where the guard rails are
Work travels along a road between two green guard rails. At each gate it is sorted: some work goes on to AI, and some is turned off into a lane that stays inside. Switch the rails off and only the sorting changes. The people stay where they are.
AI Safe Use reads how AI is really used around the work and where the guard rails are. It is never a compliance tool and never a number.
Illustration. A road runs from the bottom left to the top right between two green guard rails. Small round work tokens travel along it. Three gates stand across the road, each with a shield outline: What may go to AI, Who checks, What stays inside. Two people stand beside the start of the road as the team, and one person stands beside the second gate as the checker. With guard rails on, hollow tokens are turned off at a gate into a side lane labelled Stays inside, and the tokens that pass the checker gain a small green tick. With guard rails off, the rails and shields go grey and dashed, every token rolls straight through to the end labelled Goes to AI, and nothing is sorted. The people never change. There are no numbers or scores anywhere in this picture.
With the picture selected, press Space or Enter to switch the guard rails on or off.
- Gate oneWhat may go to AI
- Gate twoWho checks
- Gate threeWhat stays inside
- Work that may go to AI
- Work that stays inside
- Guard rail
AI Safe Use is one of the instruments AIVROs uses. It reads habits around the work and shows where the guard rails are, in words. It is never a compliance tool, never a number, never a score and never a percentage.
Illustration only. It shows the idea, not product data.
What a sponsor sees
The sponsor is the company that runs a campaign. By default, a sponsor sees cohort results: heat maps, averages, distributions and the names of the people who qualify as AI champions, without their scores. Every figure is counted on the people who actually answered, and a group of fewer than five is hidden. The check runs on our servers, so it cannot be switched off in a browser.
If the workspace owner turns on individual scores, they accept that the scores are for growth, not for hiring, firing, discipline or ranking. Members are told that their sponsor can choose to view individual scores. Culture and Climate, AI Safe Use and Role Horizon are never shown to a sponsor per person, whatever the setting.
The Leadership Fit Review is a named leadership sitting. The sponsor reads it by name, and the person is told this before starting. Named Leadership Fit Review results appear only once the workspace owner turns them on, after reading the same disclaimer, and every change is recorded.
No tracking software
AIVROs installs nothing on anyone's computer or phone. There is no browser extension, no desktop agent, no keystroke or screen capture, and no access to email, chat or code. Results come from the assessments people take and the documents a sponsor decides to upload.
AI and your data
Each campaign has an Enhance with AI control, and each person has the same control for their own results. When it is on, the text needed for a task, such as drafting a report or answering a coach question, is sent to an AI provider listed below. Sponsors' documents stay inside their own workspace and are never shared with another organisation.
We don't train on your data.
To follow your data from the browser to the database and to an AI model, read How your data flows.
How the data is protected
- Data is stored in Supabase in the United States (us-east-2). The application server runs on Render in the same region.
- Every database table has row-level security. Browsers never read member data directly; our server does, with a key the browser never sees.
- Sign-in uses httpOnly session cookies that page scripts cannot read, sent over HTTPS only in production.
- Every request is rate limited, with tighter limits on sign-in, sign-up and AI features.
- The production server refuses to start with weak or default secrets.
- Pages are served with security headers, including HSTS, a frame block and content-type protection.
How long we keep data
A roster with reporting lines uploaded to the Work Redesign Map is deleted after 180 days. A member can ask to close their account from their settings. We pass the request to their organisation, which owns the results of the assessments it gave them, and delete anything from their own AIVROs One plan. A sponsor can delete the documents they uploaded at any time. The privacy policy has the full detail.
Subprocessors
These companies process data for us, each for one job:
- Supabase: Database and sign-in.
- Vercel: Website hosting.
- Render: Application server hosting.
- OpenAI: AI drafting, when Enhance with AI is on.
- OpenRouter: Routing to AI models for some features, such as the member coach, when Enhance with AI is on.
- Stripe: Payments.
- Resend: Email, such as invitations and notifications.
- Sentry: Error reports, so we can fix what breaks.
- Gamma: Presentation decks, only when a sponsor creates one.
- Tavily and Exa: Web research on public company information.
- Google Analytics: Website visit statistics, only after you allow analytics.
Clicking a film on this site loads that film from YouTube. YouTube is a separate company, not a subprocessor working for us, and its own privacy policy applies.
Certifications
SOC 2 controls are in progress. We do not hold a SOC 2 report or an ISO certificate today, and we will say so here when that changes. Our privacy policy and terms are a draft for legal review until a lawyer signs them.
Questions and reports
For a security question or to report a problem, write to info@skillement.ai or use the contact page. See also the privacy policy and the terms.