Decision rights: the box that decides whether your AI project survives

Solution by Thanh Dao · strategy · 2 min read

Most AI projects don't break on the model. They break the first time an agent does something reasonable that nobody was willing to put their name against.

Governance is usually treated as something you assess about an organisation — once, at the top. But an agent doesn't consult the org chart. It acts, per request, thousands of times, in places no committee is watching. So the boundary has to exist as a property of the running system, decided per decision, before the build.

That's what cell AI3 (Decision rights) on the Business Solution Canvas forces: which decisions may the agent make alone, which need a human signature, and what happens to the request in between.

The uncomfortable test of whether AI3 is really filled: is the boundary in the code, or only in the prompt? A prompt is a request to a model. Code is a rule it cannot talk its way past. Those are not the same guarantee, and the difference only shows up on the day it matters.

Three examples from our own products, all deliberately deterministic:

None of that makes the AI weaker. It makes it deployable.

Try it this week: take one agent you already run, list every decision it touches, and mark each one agent alone or human approves — money, law, health and anything irreversible always in the second column. Then check where each line is actually written. If it's in a prompt, it's a preference. Move it into a gate and it becomes a boundary.

The Business Solution Canvas is free and openly licensed (CC BY-SA) — one page, no sign-up.

--- Reference — canonical BIXSO documentation: https://bixso.ai/research/decision-rights-the-box-that-decides-if-your-ai-survives

References — canonical BIXSO documentation

Work with Thanh Dao and others like them

bixso.net members publish solutions like this one and are ranked by the ones peers validate. Free to join — publish, cite, or ask for help.

Create a free account → Browse bixso.net