One workstation.
Every developer, a team.
writ turns the AI assistants you already use into a governed team of agents: a builder in every project, a shared test platform for every language and an independent reviewer. Your decisions are in front of every agent on every prompt, and enforced when one is ignored.
Your decisions, every prompt
Settled decisions restated, in your words, with every request to every agent.
One desk for the team
Every agent, project and assistant through one governed front door.
Refused, redirected, evidenced
When a decision is ignored, the action is refused, the approved way is handed back, and both are recorded.
AI agents forget what you decided.
An instruction given at the start of a session fades as the work goes on. Rules files get loaded once and left behind. Then the agent relitigates a settled decision, weakens a failing test or calls work done that isn’t.
writ doesn’t rely on the agent remembering.
of an instruction’s effect was gone after 2,048 tokens of unrelated work.
In our testing, we placed an instruction in an AI’s context, let it work on something else, then measured how much still applied.
A governed team, not a pile of chat windows.
Decisions on every prompt
Your settled decisions, restated in your words to every agent, at the workspace and machine level.
A record agents can’t rewrite
Decisions, open work, facts and handoffs, kept durably. Nothing edited or deleted.
A person approves
New tools and changes to tests wait for a person’s yes, for that exact change.
Refused, with the way through
A rule your organisation sets is enforced when the agent acts, with your approved alternative handed back.
Each developer now runs an engineering team.
A builder in every project. A shared test platform for each language. An independent reviewer that never edits code. writ keeps them on your decisions, routes requests between them, and wakes the one that owes the work.
In our own daily use: 6–18 agents at once on one workstation, across a live ~30-project estate.
Meet the teamFrom one developer to the whole organisation.
Run a team, not a chat window.
Agents across every project that keep to your decisions, wake each other when work arrives and pick up where they left off.
The same decisions everywhere.
Shared conventions in front of every agent, a reviewer that isn’t the author, and tests that can’t be argued away.
One approved policy, proven.
Security, platform and compliance teams publish policy to every seat and see where it’s enforced, with evidence for audit.
Governs the assistants you already license.
writ is not another coding agent. It sits in front of the ones your developers already use, in the editors they already use.
Built under real load.
Figures are from Verilex’s own daily use, not a customer result.
See a team at work, and a refusal.
A live demo on your own workflow. Technical deep dive available under NDA.