Who it’s for
Every organisation has the same problem: the knowledge of what you can and can’t do is scattered across people, teams, and functions. This was already expensive before AI. Now it’s ungovernable.
Governance is already fragmented. AI just exposes it.
A new initiative touches legal, comms, procurement, finance, and operations. The constraints that govern what the organisation can commit to live in different people’s heads, across different teams, in different documents nobody reads at the same time. Even without AI, figuring out “who knows what” and “who’s responsible for what” is the actual work of governance.
Right now, this means everything grinds to a halt. An initiative that should take days takes weeks or months — waiting for legal to review, then comms to sign off, then procurement to clear it, then someone’s on leave and the whole chain stalls. AI makes the work faster but doesn’t fix the coordination bottleneck — whether it’s an AI agent acting autonomously or a person using AI tools to move faster. You still need five people in three departments to say yes.
The failure isn’t speed. It’s that institutional constraints are distributed across functions, and there’s no infrastructure that holds them together.
Two ends of the spectrum. Same problem.
Scenario A
A public authority with 40 pieces of legislation
A government agency uses AI to help caseworkers process citizen requests. Employment law, privacy regulations, FOI requirements, procurement rules, ministerial directives — 40 legislative instruments across a dozen teams govern what the agency can and cannot commit to.
A caseworker uses Claude to draft a response. It crosses into procurement territory, triggers an FOI obligation, and contradicts a ministerial directive from a different department. The caseworker doesn’t catch it — not because they’re careless, but because those constraints belong to other people in other teams. The knowledge was never in one place. This was true before AI, but now responses go out in minutes instead of days.
Today
Everything grinds to a halt. Each response needs sign-off from legal, privacy, procurement. Weeks of back-and-forth. Or — more likely — people skip the coordination and hope for the best. Compliance reconstructs the damage in a quarterly review.
With Constellation
Legal’s constraints, procurement’s thresholds, privacy’s rules, and comms’ policies are all encoded as constraints. Every action is checked against all of them — simultaneously, instantly. No coordination chain. Move at the speed of the machine.
Scenario B
A three-person startup shipping an AI agent
A founder, an engineer, and a designer. The founder is simultaneously legal, comms, product, and sales. Their AI agent handles customer onboarding — sends emails, answers questions, schedules demos, responds to support tickets.
The agent told a customer the product supports SSO — that’s a product constraint. It quoted an enterprise pricing tier that hasn’t been approved — that’s a finance constraint. And it promised data residency that would require infrastructure changes — that’s an engineering constraint. Three different “departments” worth of knowledge, all in the founder’s head, none of which they were thinking about when they skimmed 40 outgoing messages between meetings.
Today
They find out when the customer asks for SSO, quotes the price they were promised, and asks about data residency. Three commitments, three problems, all already made. The founder can’t wear all the hats at once.
With Constellation
Three constraints — “don’t commit to unshipped features,” “only quote approved pricing,” “no infrastructure guarantees without engineering sign-off” — fire before the messages send. The founder’s knowledge across all their hats is encoded once and enforced at machine speed.
The pattern is the same at every scale: institutional knowledge is fragmented across people and functions. Today that means everything either grinds to a halt waiting for coordination, or moves fast without confidence. Constellation gives you both — speed and the structural confidence that governance is already in the architecture.
40%
Faster decisions
100%
Audit compliance
18%
Lower coordination cost
0
Silent authority gaps
Constellation is not for
Internal productivity tools
If your AI only helps with internal tasks that never cross institutional boundaries — summarising notes, formatting documents — you don’t need governance at the moment of action.
Growth optimisation
Constellation does not optimise outcomes. It enforces institutional commitments.
Advisory AI
Systems that recommend but never act don’t need governance at the moment of action. They need good judgment from the humans who do act.
Surveillance or performance scoring
Constellation is constitutionally prohibited from ranking individuals, scoring compliance, or enabling trace-based discipline. If that’s what you want, this is not your product.
If actions taken on behalf of your institution — by people, AI, or both — can commit you to something, we should talk.