Questions
Why do I need this?
Real people are often affected by what an AI system produces, and there can be legal consequences when it goes wrong. Building responsible, trustworthy systems is good practice, and helps to establish and protect your company’s reputation, so it has real economic value.
Yet usually nobody has written down who is accountable for what in an AI supply chain. The company in the middle, the one that added an AI feature to a product it already sold, often finds out only when a customer asks. Boundsmap writes it down: who stands where, which obligations are yours, which are your supplier’s or your customer’s, and what you need from each. Then you can answer, and see the gaps, before someone asks.
What is the idea behind it?
An AI product is rarely made by one party. Boundsmap’s Roles & Boundaries Framework places everyone involved on five layers, from the people who make the model to the people it affects. It then looks at the four boundaries between them, where one party hands something to the next and has to be trusted about it. Many obligations sit at those handovers, so that is where the map shows who is accountable.
The framework began in PhD research at Cardiff University, and was tested with practitioners from industry, government and academia.
Why not just ask AI?
You can, and AI is a good way to learn about the law. The difference is what you get back. Ask AI the same question twice and the answer can change. It can also cite an article that says something else, or one that does not exist, in the same confident tone.
In Boundsmap, our AI does one job: it reads your description and works out who is involved, and you confirm or correct that. Then fixed rules that we write and test, not our AI, decide what the law asks at each handover. The same facts give the same answer. Every obligation names the article it comes from, so you or your adviser can check it. And where the answer depends on something only you know, it says “Needs an answer” instead of guessing.
Is this legal advice?
No. Boundsmap works from your description and your answers. It is not an audit, a certification or legal advice, and it does not replace an adviser. What it gives you is a clear, checkable map of your own position to take to your team, your customers and a lawyer. Take legal advice before you rely on it.
Which laws does it cover?
The EU AI Act and UK and EU data protection law. If a customer requires the NIST AI Risk Management Framework (NIST AI RMF), a framework from NIST, the US standards body, Boundsmap also places its four functions on your map. That is not law, and Boundsmap does not assess you against it. A country Boundsmap has no rules for is marked “Not covered”.
What do “Needs an answer” and “Not settled” mean?
“Needs an answer” means the map needs a fact only you know before it can say what applies to you, and it says which question answers it. “Not settled” means the law itself has not decided, and nothing you could say would change that. The map lays out the readings and assigns nothing.
What does your AI see of my description?
Your description goes to Claude, Anthropic’s AI model, which works out who is involved. Your maps are stored in our database, run for us by Neon, in London. The privacy notice says what goes where, and for how long.
Do I need a compliance person to use it?
No. Boundsmap is made for people who do not have one: a founder, an ops lead or a product manager with an AI feature in a product. It says in plain words who is accountable at each handover and what the law asks, with the article beside it, so you can state your own position and answer your customers’ questions. It is not legal advice, so check it with an adviser before you rely on it.
What does it cost?
Your first map is free. Pro is £10 a month and Team £39 a month, at launch prices. See the plans.
There is more in the guides, or use the contact link below.