Foundry is the AI engineering platform for your whole team. It’s a desktop app for running AI agents on real software: you describe the work, and Foundry breaks it into stories. Agents build each story in parallel, your tests run again, and a second model reviews the work. You approve the change before it ships.
Your team is running more work than anyone can keep in their head. Foundry keeps track of all of it, and makes sure everything you ship is built the same way. This post answers the questions we hear most about how it works.
Your team is running more work than anyone can keep in their head. Foundry keeps track of all of it, and makes sure everything you ship is built the same way. This post answers the questions we hear most about how it works.
Who makes Foundry?
Engineering11 is the company behind Foundry. We build the tools, then we build with them: you can use Foundry with your own team, or have our team build your software with it.
How does Foundry run the work?
With the same steps, every run:
Plan. You describe the work, and Foundry breaks it into stories.
Build. Agents build each story in parallel.
Check. Your tests run again, and a second model reviews the work.
Approve. You approve the change before it ships.
A controller hands each story to agents running in parallel, each in its own workspace, and the agents can be sandboxed if you want. Agent sessions, Git, and a terminal are all in one window, where your team works.
Plan. You describe the work, and Foundry breaks it into stories.
Build. Agents build each story in parallel.
Check. Your tests run again, and a second model reviews the work.
Approve. You approve the change before it ships.
A controller hands each story to agents running in parallel, each in its own workspace, and the agents can be sandboxed if you want. Agent sessions, Git, and a terminal are all in one window, where your team works.
Why can’t an agent mark its own work as done?
Because an agent saying it’s done never counts. Foundry checks the work itself: it reruns your tests, types, lint, and security checks, and records the results as evidence for each story.
Then a different model reviews each change against your docs. One model builds, and another reviews, and its findings are fixed before a person looks. After that, you approve it.
Then a different model reviews each change against your docs. One model builds, and another reviews, and its findings are fixed before a person looks. After that, you approve it.
What is the Knowledge Center?
Agents read your patterns before they write. Foundry builds a Knowledge Center from your code: your architecture, patterns, and decisions, each with an owner. Every agent reads it before it writes, and every doc is checked like code before agents can use it.
Foundry keeps the Knowledge Center matched to the code as it changes. When the code changes, Foundry flags the docs that are out of date and drafts the updates for your team to review.
Foundry keeps the Knowledge Center matched to the code as it changes. When the code changes, Foundry flags the docs that are out of date and drafts the updates for your team to review.
Why does consistency matter so much?
Because every service should read like one team wrote it. With Foundry, errors, logging, permissions, and data access work the same way everywhere, so people and agents get it right the first time.
Foundry keeps the science consistent, so your team can focus on the art. Every change follows the same steps and the same patterns.
Foundry keeps the science consistent, so your team can focus on the art. Every change follows the same steps and the same patterns.
Which models does Foundry use, and does it work with existing code?
Your own. You run Foundry on your own code, with your own models. And yes, Foundry works on the code you already have.
For new products and replatforms, there’s an optional add-on: the Full-Stack Foundation. Agents build a complete, working platform into your repo, documented end to end, with authentication, permissions, payments, and notifications done before sprint one.
For new products and replatforms, there’s an optional add-on: the Full-Stack Foundation. Agents build a complete, working platform into your repo, documented end to end, with authentication, permissions, payments, and notifications done before sprint one.
Does Foundry replace engineers?
No. Your team decides what to build, approves every change before it ships, and puts its craft into the product. Foundry handles the part that should be the same every time.
Do you use Foundry yourselves?
Every day. After its first release, Foundry built the rest of Foundry, and our engineers use it to build software for clients. It’s the same chain we offer you: Foundry, the Knowledge Center, the Full-Stack Foundation, and then the client’s software.
How do I get started?
Fill in the short form on our contact page, and we’ll set up a walkthrough with your team, on work like yours. From there, start with Foundry, or have our team build it. More answers are in the FAQ.