In a regulated enterprise, the committee is the AI bottleneck
Code generation is solved. In a regulated enterprise the constraint is the approval process, so the fastest path to adoption is one that needs no new tool approved.
Written by
-
Benedikt Stemmildt Founder & Co-CEO
Published: 4 August 2026 Updated: 1 min read
A large insurer wanted the factory. Special budget approved. Board sponsorship. Architects who could not wait to start. Development teams standing by.
Then someone asked the only question that mattered: how long does it take to get a new cloud tool approved? Six to nine months. Thirteen regulatory committees, four of them just for cloud. A works council with its own co-determination rights. The budget was real. The enthusiasm was real. None of it touched the timeline.
The constraint is the committee
Code generation is solved. The model writes the code. So the constraint is not the technology. It is the committee standing between the team and the tool.
That reframes the whole adoption problem. Most plans start with picking the best AI stack and then walk it through approval. In a regulated enterprise that walk takes the better part of a year, and the year is the project. By the time the new gateway clears the last committee, the budget cycle has turned and the architects have moved on.
Start with what is already approved
So flip it. Do not start with the best tool. Start with what is already approved.
- The cloud platform that cleared compliance two years ago already runs a model service.
- An identity provider already in production already does the auth.
- A thin gateway in front, self-hosted on infrastructure that is already signed off, routes the calls.
Same capability the new stack would have given you. Zero new tool approvals.
The gate sits upstream
The same shape shows up elsewhere. Another enterprise budgeted real money per month for AI inference and spent under a tenth of it, because the constraint was never budget. It was calendar capacity, and protected time was a leadership decision nobody had made.
Every adoption gate sits upstream of where leaders look first. Before you pick the tool, find the committee.
Where is your AI rollout actually stuck, and is it a technology problem or an approval one?
Further reading
Written by
-
Benedikt Stemmildt · Founder & Co-CEO
More on Agent security and risk
- A hook is not a boundary: why agents need a sandbox A training group got an agent past a hook that guarded a secrets file. A hook filters the calls it was written for, a sandbox limits what the agent can reach, and the trade between the two is a company decision.
- Every automation needs a scoped, reversible sandbox Teams design what their automations do and neglect where they run. A written blast radius, a narrow scope and throwaway environments decide whether you can leave an automation unattended.
- Vouch for the harness, not for its output Code review cannot keep up with agent output. Open source maintainers already admit contributors instead of reviewing contributions, and in an agentic factory the contributor to admit is the harness.
Test the constraint in your system
The Factory Readiness Assessment connects this research to your value stream, your data, and the next decision.
Training on this topic: Agentic Engineering Security for Development TeamsWant to check one team first? The self-check: twelve yes or no questions, three minutes, instant result, no account. Take the self-check