Open Source
The tools we build with are public. Read them before you hire us.
Most agencies ask you to trust the process. We would rather you read it. The gate that governs how AI touches our work is open source, MIT-licensed, and on GitHub now. Your engineers can audit it this afternoon, before a contract exists.
Why a shop open-sources the tools that govern its own AI
A claim you can check beats one you're asked to believe.
An agency that says its AI is “human-supervised” is asking for faith. One that hands you the code enforcing the supervision is handing you something you can test. The second is harder to fake.
Readable code is one less thing for your compliance team to take on trust.
Software you can read has no vendor breach to inherit and no per-seat license that turns your own data into someone else's recurring revenue. The audit question stops being “do we trust this vendor” and becomes “does this code do what it says,” a question your own people can answer.
Greenroom · open source · MIT
The sign-off gate, in code you can read
Greenroom is per-story review and sign-off for Storybook. A person reviews the rendered component, pins a comment to the exact spot, and approves it. Our AI coding agents read that feedback over MCP, with the story, the file, the selector, and a screenshot of the pin, and do the work. Nothing ships until a human signs off.
It is the smallest honest version of the rule we build by: AI can write the change and even apply the fix, but it cannot approve its own work. Every sign-off is a person, on the record. The sidecar that runs the review contains no AI of its own. The agent is external, and by construction it cannot sign.
Read Greenroom on GitHub# github.com/igility/greenroom
- MIT-licensed, public, CI-green. Pre-release and under active development: read it, run it, open an issue.
- A Storybook addon, a review sidecar, and an MCP server. The human reviews rendered components; the agent reads the sign-off.
- Approval is human-authorized and audit-labeled, so the gate can be checked.
Acorn · self-hosted
The analytics your data never leaves
Acorn is the product and marketing analytics behind Igility builds. It is self-hosted, so it runs on infrastructure you own, and the behavioral data it collects, including anything sensitive, never touches a third-party analytics vendor's servers. For a healthcare or regulated client, that is the difference between “we have a BAA with our analytics provider” and “there is no third-party analytics provider.”
It's the honest version of the sovereignty pitch: when we say your data stays yours, self-hosting is what makes it true.
Acorn is the next tool we're opening. It runs self-hosted today; the source goes public next, so your engineers will be able to read it the way they can read Greenroom now.
The rule under all of it
One line governs everything we ship: no model output reaches a user or a system of record without a deterministic gate. We build with AI and retire the risk at review. We ship software your own agents can operate through constrained interfaces, on your terms. We put a model inside the running product only as a deliberate, eyes-open client decision, never as a default.
Greenroom is that rule made checkable for the work humans need to see with their own eyes. Read the three postures in full, or read the code. Both say the same thing.
The code is public. The next build could be yours.
Start with the repo and read how we gate AI before it ships. When you are ready, book a working call and we will scope the build.