Resources
How Foundry actually operates.
Everything here explains the mechanism rather than the pitch. If you are trying to work out whether you could trust an AI product team with real work, start with the authority model.
Guides
Start here.
The operating model
How a business request becomes product work, moves between agents, and reaches a release.
The AI product team
What each of the eight roles owns, does, and hands to whom.
The authority model
Which actions are autonomous, which need your approval, and which no agent may ever hold.
Tool permissions
Why connecting a system and granting authority over it are two separate decisions.
Inside a product workspace
The work queue, the decision queue, the activity log and the release view.
Integrations and availability
What Foundry supports today, what is planned, and what is not built yet.
No case studies yet. Foundry is early, and we would rather have an empty shelf than a fabricated one. When there are customer results worth publishing, they will appear here with the customer’s name on them.
Questions
What businesses ask us first.
- Is Foundry a software development agency?
- No. Foundry is a B2B SaaS product. You deploy AI product teams through software you operate — there is no engagement to scope, no contract per project, and the product context stays in your workspace rather than walking out with a delivery team.
- Do I need developers to use Foundry?
- Not for basic product workflows. You describe what the business needs and approve the consequential decisions. Technical teams get more out of it — Foundry runs alongside an existing engineering organisation and takes defined workstreams — but it does not require one.
- Can Foundry work with an existing codebase?
- Yes, once a repository integration is connected. Foundry reads the repository as technical context before it plans anything, so the work fits what is already there rather than proposing a rewrite.
- Does Foundry deploy directly to production?
- Only within the permissions and approval rules you configure. By default, production deployment is approval-required: DevOps prepares the release, reports what passed, and stops until a human approves it.
- Can Foundry replace my entire product team?
- That is not how we would position it. Foundry is designed to perform an increasing amount of repeatable digital product work, while your business retains strategic and consequential decision authority. Some of what a product team does is judgement about your market, your customers and your commitments — that stays with you by design, not by limitation.
- What stops an agent doing something it should not?
- Permissions are enforced in backend logic, not in a system prompt. Every action is classified as autonomous, approval-required or human-only, and every tool carries its own separate read, create, modify, execute and delete grants. An agent cannot widen its own authority — that action is human-only.
- How do I know what Foundry actually did?
- Every action is recorded: which agent, what it did, why, which tools it used, what changed, whether approval was required, who approved it, and what the result was. You can inspect the whole history of the work, not just its current state.
- What happens when an agent gets it wrong?
- QA sends it back. A feature is complete only when its acceptance criteria pass, and QA — not the engineer who wrote it — decides that. Work that fails returns to the responsible agent with the failing criterion attached.
Start with one product
Tell Foundry what your business needs.
Create a workspace, describe the product, and watch the team pick it up. You approve what matters; Foundry handles the rest.