Your AI Product Team
Tell Foundry what your business needs. A coordinated AI product team plans it, designs it, builds it, tests it and ships it — and comes to you for the decisions that are yours.
Foundry does not give your business another tool for building software. It gives you a product team that does the work.
Release readiness 82% — one decision is yours
Current objective
Launch customer ordering for the Nigerian market
Your team right now
Checkout requirements completed
12 user stories created
Checkout flow designed
6 screens ready
Payments API completed
Contract published to Frontend
Integrating checkout
7 of 11 tasks complete
Testing payment flow
3 issues found · 2 fixed
Needs your decision
Foundry recommends Paystack. It settles locally in naira and needs no additional entity to onboard, and the payment adapter stays provider-agnostic either way.
01
Plan
Requirements and user stories
02
Design
Flows and screens from the plan
03
Build & test
Engineers and QA, in step
04
Ship
Releases you approve
Product in action
Not a prompt returning text. A business request enters the team, each agent takes the part it owns, and work moves between them until something ships.
“Build table ordering for our restaurants.”
Defines the V1 objective
A diner orders and pays without waiting for staff
Creates the scope
12 stories · 34 acceptance criteria
Designs the ordering flow
6 screens · every failure state covered
Builds the APIs and publishes the contract
Builds the interface against that contract
Runs the acceptance checks
28 checks · 3 issues found
2 fixes completed and returned
Each issue went back to the agent responsible, not to you
All acceptance criteria pass
28 of 28 checks passed
Prepares release 1.4
Healthy on staging · no blocking defects
Production approval required
DevOps stops here — deploying is your call
Shipped
Live across 22 locations
How Foundry works
Describe the product, the problem or the feature in your own words. Add whatever context you already have.
“We need a customer portal where businesses can book deliveries, track shipments and download invoices.”
Foundry runs discovery against your context, defines the scope that serves the objective, and turns it into work with acceptance criteria.
24 work items created
Design, frontend, backend and QA work the same plan at once — assigning to each other, and blocking openly when something is missing.
Foundry runs the routine execution on its own and escalates the consequential calls. Everything else keeps moving while it waits.
Comes to you
Meet the AI product team
You brief the Product Lead. It decides what the objective needs, and the rest of the team picks up the parts they own — including handing work to each other.
Product
Engineering
Delivery
These are not eight chatbots you brief individually. Select a role to see what it owns.
Owns the outcome and coordinates the team.
Does the work
Works with the team
Receives the objective from the business, directs every other agent, and is the one agent that asks you for a decision.
Multi-agent collaboration
This is what separates a product team from a set of assistants. An engineer that needs something blocks and asks for it. QA sends work back when it does not meet the criteria. Nobody marks their own homework.
Work item
Password reset
Password reset implementation
6 acceptance criteria, including a 30 minute link expiry
Reset endpoint missing
Blocked itself rather than stubbing a contract that does not exist
Reset endpoint created
Single-use tokens, rate limited. Contract published
4 screens integrated
Token expiry test failed
A token issued 45 minutes ago still authenticated
Expiry now evaluated server-side
Cause was the client clock
12 of 12 tests passed
Release ready — needs your approval
Approved production deployment
Release 1.4 live
Every handover is a structured operation against a shared work item — ASSIGN, REQUEST, RETURN, ESCALATE — not a message in a chat window. That is what makes the work inspectable afterwards.
The product
This is Acme Foods’ checkout — the work you watched move through the team above. The Designer specified it, the Frontend built it against a published API contract, and QA held it until every acceptance criterion passed.
Your order
Ikeja · DeliveryDelivery details
Payment
Confirmation state: estimated delivery window
Specification, not decoration: every field, every state and what the screen does when payment fails.
Your order
Ikeja · DeliveryDelivery details
Payment
Arriving today, 6:40–7:10pm
Built to the spec and wired to the contract Backend published — not a redraw of the mockup.
Three issues were found during this cycle. Each went back to the agent responsible and was verified before the release moved.
DevOps prepared the candidate and stopped. Production deployment is approval-required — a human released this.
Acme Foods is a sample company used to demonstrate Foundry. This is not a real customer’s product.
Product lifecycle
Shipping is the middle of the loop, not the end of it. What customers actually do after release becomes the next round of product work.
Idea to release
Learns the business and the problem
Defines the objective and the work
Flows, screens and failure states
Frontend and backend against the spec
Verified against acceptance criteria
Released once you have approved it
After release
Launch, campaigns and experiments
What customers actually did
Turned back into product work
Back intoPlan
Why Foundry is different
Plenty of tools make a developer faster. Foundry is aimed one level up — at the work a product organisation does, and at who is accountable for it.
AI coding tools
Help a developer write code.
The developer still owns the problem, the plan, the design, the testing and the release.
Foundry
Take responsibility for the product work.
Including every part of it that is not code.
AI chat assistants
Answer a question and return text.
What happens next is entirely on you.
Foundry
Turn a request into assigned work.
Carried to a verified result by the agent that owns it.
Software agencies
People you brief, contract and wait on.
Capacity is headcount, and the knowledge leaves when the engagement does.
Foundry
Software you operate.
The context stays in your workspace and the work is auditable while it happens.
Project management tools
Track the work that humans do.
A record of the work, not the worker.
Foundry
Do the work.
The tracking is a by-product of the team doing it.
Use cases
Foundry suits any team with more product work than product capacity — which, in practice, is most of them.
Startups
You have the insight and the customers. What you do not have is six months and a founding engineering team.
SMEs
The work that never justifies a hire — a customer portal, an ordering platform, the automation that kills a weekly spreadsheet.
Growing companies
Your product needs more than your team can deliver this quarter. Foundry runs defined workstreams alongside the engineers you have.
Enterprise product teams
Scoped, bounded and auditable. Each workspace has its own permissions, approval rules and complete record of every action taken.
Internal business teams
Operations, HR and finance tools rarely win the priority argument against customer-facing work. Foundry gives them a team of their own.
Inside the product
Not a dashboard of charts. A view of what the team did, what it is doing, what is blocked, and what is waiting on you.
Stage: BUILD · Release 1.4
Working now
Ask Foundry
“Add a referral system before launch.”
5 work items created and assigned
The reward value is a spend commitment, so it will come back to you as a decision before anything is built against it.
Needs your decision
Foundry recommends Paystack for the Nigerian launch.
Latest activity
Foundry started the guest checkout regression sweep
Running 28 acceptance tests against staging build 1.4.0.
7 Sept, 06:41 · AI QA Engineer
Foundry built the checkout payment step
Implemented the payment step against the provider-agnostic contract. Two acceptance criteria still failing.
7 Sept, 06:38 · AI Frontend Engineer
Foundry drafted order status notification requirements
Three notification points identified from support themes: accepted, preparing, ready.
7 Sept, 06:20 · AI Product Manager
Conversation is a command surface — what you ask for becomes assigned work with an audit trail.
The authority model
A DevOps agent with cloud access does not thereby have permission to provision unlimited infrastructure. A Growth agent with your social credentials does not thereby have permission to publish. Every permission is explicit, and enforced in the backend — not asked for politely in a prompt.
Done without asking. Nothing here reaches a customer or spends your money.
Reads context. Changes nothing.
Proposals you can reject at review.
Nothing reaches a customer until it is built and approved.
Lands on a branch, not on production.
Read-only against non-production environments.
Scoped to an open defect with acceptance criteria.
Internal artefact.
The work is ready and the recommendation is stated. Nothing proceeds until you decide.
Changes what you are paying for and when it lands.
Expensive to reverse once built on.
Reaches your customers.
Data loss is not recoverable by retrying.
Spends your money. Cloud access is not a budget.
Widens the blast radius of every later action.
Speaks publicly in your name.
A commercial decision, not a product one.
No agent holds these at any permission level, and none can grant them to itself.
An agent must never be able to expand its own authority.
Reserved. No agent holds this, at any permission level.
Binds your company. Not delegable.
Protects the approval model itself.
Security & trust
Giving software agency over your product only works if the boundaries are real. These are enforced in the platform, not requested of the model.
Product context, code and conversation never cross between organisations.
Permissions are checked in application logic before an action runs.
Every action is recorded with its reason, its approver and its result.
Credentials are referenced, never held in context or passed to a model.
On certifications. Foundry is early. We are not going to display a compliance badge we have not earned. If you need a specific control framework in place before you can adopt Foundry, tell us which one and we will be straight with you about where we are.
Integrations
Connecting a system lets Foundry read it. What Foundry may then do with it is a separate, explicit grant — read access to a repository is not permission to merge.
Design context
AvailableCode and repositories
AvailableProgress and approvals
AvailableInfrastructure and deployment
AvailablePlus 10 more across planning, cloud, insight and business systems. See what Foundry supports
Availability describes what Foundry supports, not what is connected to your workspace. Nothing is connected until you connect it. Marks are Foundry’s own — no third-party branding is used.
Business outcomes
Foundry is new, and we are not going to quote you savings percentages or customer counts we do not have. Here is what the operating model actually changes.
The handovers that normally need a standup, a ticket and three follow-ups happen between agents, against a shared work item.
Design does not wait on a review slot. Frontend does not wait on a sprint boundary to get an endpoint.
Workstreams you shelved because nobody had the bandwidth become work you can actually start.
The team does not stop between your working hours, and picks up where it left off with its context intact.
Every work item has one owning agent, one requesting agent, and a record of who moved it and why.
Foundry acts where you have granted authority and stops where you have not. There is no implicit escalation.
Pricing
Plan structure is set. Commercial values are not published yet, so we are not going to print numbers that will change — talk to us and we will give you real ones.
One product, getting it built.
Contact sales
Several products in flight at once.
Contact sales
Multiple teams, production workflows.
Contact sales
Bounded workstreams under your controls.
Custom
Questions
More questions answered — including how permissions are enforced and what happens when an agent gets it wrong.
Start with one product
Create a workspace, describe the product, and watch the team pick it up. You approve what matters; Foundry handles the rest.