Widal.Factory
Tickets in. Reviewed pull requests out. In your own cloud.
Widal.Factory hands your tickets to AI coding agents. Each agent opens a pull request, and your engineers review and merge it. It runs on your cloud infrastructure, and we operate it, so you keep your code, your data, and control over our access.

The path of one ticket
- 01TicketEngineer
Labels a ticket in Linear or GitHub Issues. That is the hand-off.
- 02BriefFactory
Makes a git worktree and branch, and starts one coding agent with the ticket, the repo and your rules.
- 03QuestionAgent
Asks on the ticket when something is unclear, and waits for the answer instead of guessing.
- 04Pull requestAgent
Pushes the branch and opens a pull request. Your CI runs on it like on any other.
- 05ReviewEngineer
Reviews the pull request. Comments go back to the same agent, which fixes and pushes again.
- 06MergeEngineer
Merges. The factory closes the ticket, cleans up, and records time and cost.
The agents
- Read the ticket and the code
- Write the change and the tests
- Open the pull request and answer review comments
- Ask when the ticket is unclear
Your engineers
- Decide which tickets go to an agent
- Answer the agents' questions
- Review every pull request
- Merge. Nothing reaches your main branch without a person
We run it. You own it.
On your cloud infrastructure
The factory, its data, its backups and its encryption keys live in your own AWS account. We deploy, update and monitor it there.
Our access is yours to control
We work through roles you create from a template you can read first. Every action shows in your cloud audit log. You can remove our access at any time.
Your model account
Agents use Claude through your own cloud or Anthropic account, under your agreements. You pay the model provider directly, with no markup from us.
A factory per engineer, one hub
Each engineer gets their own factory in its own virtual machine, so one person's agents never touch another's work. Every factory reports to one hub: what agents are doing, what waits for a person, and what each merged change cost.
Questions we get
- What does it cost?
- We give prices on a call, once we know your team and the work. A named Widal engineer is part of every engagement, for setup, new workflows and the weekly review. Model use is billed by your own cloud or Anthropic account, with no markup from us.
- Can agents merge on their own?
- No. Every pull request needs a person to review and merge it. Agents hold no GitHub key of their own, and with your main branch protected, nothing reaches it without a person.
- Does our code leave our account?
- The code stays in your cloud account and your GitHub. The agents send code to the model you choose, through your own cloud or Anthropic account.
- Which trackers and agents does it work with?
- Linear and GitHub Issues. The coding agent is Claude Code. Our engineers connect other tools during setup. The full list is under How it works, Integrations.