Add a Product Pod.
Your Product Pod is a governed AI product team inside ProductOS. Bring one customer signal and leave with a production-ready iteration, with your product owner directing the work and approving what moves.
The sprint request goes to our contact form. Tell us the customer signal you want to act on and we scope the pod with you.
Or start on FreeOne unit of work
One customer signal in. One production-ready iteration out.
A Product Pod is not a platform migration and not a new reporting line. It is a single unit of work that starts with something a customer actually told you and ends with code you can review, approve and merge.
You bring
One customer signal
A support theme, a churn reason, a sales objection, an interview quote, a metric that moved. One signal, stated plainly, with whatever context you already have.
The pod runs
The full lifecycle, in one context
Research checks the signal against evidence. The spec is written from that research. Design comes from the spec. Code comes from the design. QA drives the real product. Nothing restarts from a summary.
You leave with
A production-ready iteration
A specced, designed, built and tested change in your own GitHub repository, with the research and decisions behind it attached, waiting for your review and approval.
How a pod is organized
A person in charge, with execution underneath.
A Product Pod is not a swarm and not a black box. One human owns the direction, the ProductOS Agent plans and routes, specialists do the craft work, and everything that ships returns to the person who owns it.
Human
Product Owner
Sets direction, chooses what the pod picks up, approves each stage and stays accountable for the outcome.
Coordination
ProductOS Agent
Your orchestrator
Plans each stage, maintains the shared product context, routes work to the right specialist and returns stage decisions to the Product Owner for approval.
Specialists
Ship
Deploy Agent
Runs the preflight build, pushes to your GitHub repository and takes the deployment to your hosting and domain.
Product Intelligence
The future closed-loop layer: watching what shipped and returning it to the Product Owner as the next input. Not available today, and not part of what you buy today. Your team runs this step.
What is learned returns to the Product Owner, who decides what the pod picks up next.
The lifecycle
A loop, not a relay.
Every stage reads from the same project memory, so the spec is written against the research that was run and the build follows the design that was approved.
- 01
Signal
One customer signal enters the pod as the unit of work, with the context your team already has.
- 02
Plan
The ProductOS Agent turns the input into a stage plan and routes each piece to a specialist.
- 03
Research
Evidence, competitors and sources are gathered, logged and attached to the project.
- 04
Spec
Requirements are written section by section against that evidence, not from memory.
- 05
Design
Flows, screens and the design system are drawn from the spec that was just approved.
- 06
Build
Production code is written in the project sandbox, against your stack and conventions.
- 07
Test
QA drives the real browser and the real API, then reports severity-ranked findings.
- 08
Ship
Preflight build, push to your GitHub repository, deploy to your hosting.
- 09
Observe
Your team watches what shipped against the signal that started the iteration.
- 10
Learn
What you learn becomes the next input, and a human decides what the pod picks up next.
Learn becomes the next signal, and the loop runs again. Observe and Learn are run by your team today. The Product Intelligence Agent that would close this loop automatically is on the roadmap.
Decision rights
Who decides, and who does the work.
Agents handle the heavy execution. Experienced humans keep strategy, judgment, review and accountability. Stage transitions are proposed for approval, never taken on their own.
Your people decide
- Strategy, positioning and what the product is for
- Which signal or brief the pod picks up next
- Scope, trade-offs and what is deliberately not built
- Brand, taste and the calls that need judgment
- Whether a stage moves forward, and what merges
- Customer, client and stakeholder relationships
- Accountability for the outcome
ProductOS executes
- Research sweeps, source logging and competitor scans
- Drafting the spec section by section from that evidence
- Design system, flows and screens from the approved spec
- Production code written in the project sandbox
- QA passes across the browser, the API and accessibility
- Preflight build, GitHub push and deployment runs
- Keeping one project memory current across every stage
What changes for the team
The gap between signal and shipped gets shorter.
Not because anyone works longer, but because the work stops being handed between people who each start again from a document someone else wrote.
Signal to shipped stays continuous
The signal that started the work is still attached when the code is reviewed. Research, spec, design and build read from the same project memory instead of a fresh summary at each step.
One shared context, not five documents
The PRD is written against the research that was actually run. The design is drawn from that PRD. The build follows that design. Every artifact stays in one project wiki your team can read.
Fewer handoffs to manage
The ProductOS Agent routes each stage to the right specialist and asks you when a decision genuinely needs a human. Your product owner directs the pod rather than chasing it.
More experiments in flight
Team capacity covers 10 active products and 8 concurrent AI agents in one workspace, so more than one bet can move at once instead of queueing behind a single squad.
Code your team owns
Every product syncs to your own private GitHub repository and can be exported. Your engineers read it, review it and merge it like any other branch.
Human approval at every gate
Stage transitions are proposed, not taken. Nothing designs past an unapproved spec and nothing ships past an unapproved build.
Current capacity
What a pod runs on today.
A Product Pod runs on the capacity your workspace already includes. These are the current founding prices and allowances, published in full on the pricing page.
Team
Start hereFor startups and in-house product teams.
$599per month, founding price
- active products
- 10
- User Licenses
- 5
- Collaborator Licenses
- 10
- monthly AI credits
- 15,000
- concurrent AI agents
- 8
Agency
For agencies and product studios.
$1,499per month, founding price
- active products
- 40
- User Licenses
- 15
- Collaborator Licenses
- 30
- monthly AI credits
- 40,000
- concurrent AI agents
- 25
Every plan starts with a 7-day free trial. Allowances and founding prices are the ones published on the pricing page.
Ownership and governance
Proof, not promises
The things a product team will be asked about in the first review meeting, answered before you get there.
You own the code
Every product syncs to your own private GitHub repository and can be exported. The codebase is yours whether or not you stay on ProductOS, and its history follows a fork, a clone or a transfer.
Approval is a gate, not a formality
The ProductOS Agent proposes stage transitions and waits. Agents ask a bounded question when a decision needs a human instead of guessing and moving on.
Stakeholders can review without a seat
Review Licenses are unlimited on every plan. Anyone you invite can view, comment on and approve a product without consuming a paid license.
Permissions and budgets are yours to set
Team permissions, approval workflows, per-product AI budgets and member-level usage analytics are part of the Team workspace, so spend and access stay governed.
Questions worth asking first
What is a Product Pod?
A Product Pod is your governed AI product team inside ProductOS: a human Product Owner directs the ProductOS Agent, your orchestrator, which routes work to specialist agents for research, requirements, architecture, design, build, QA and deployment. It sits inside your existing team rather than beside it, so there is no new reporting line and no new organizational layer to manage.
Does this replace our engineers, designers or product managers?
No. Agents handle the heavy execution: research sweeps, spec drafting, design generation, writing code in a sandbox, QA passes and deployment runs. Experienced humans keep strategy, judgment, scope, taste, review and accountability. A pod without a strong product owner produces work nobody wants to merge.
What do we actually get out of a Product Pod Sprint?
You bring one customer signal. The pod runs it through plan, research, spec, design, build and test, and you leave with a production-ready iteration in your own GitHub repository, with the research and the decisions behind it attached for review.
Who approves what?
Your people do. Stage transitions are proposed to a human and wait for approval. Scope, trade-offs, brand calls, what merges and what ships all stay with your team. ProductOS executes inside those decisions and records what it did.
Do we own the code?
Yes. Every product syncs to your own private GitHub repository and can be exported at any time. Deployment goes to your hosting and your custom domain, and a product's history follows it across a fork, a clone, a new repository or a transfer to another workspace.
Is the Product Intelligence Agent available today?
No. The Product Intelligence Agent that would close the loop automatically, watching what shipped and feeding it back as the next input, is on the roadmap and is not part of what you buy today. Right now your team runs the observe and learn steps, and the pod picks up what you decide to act on next.
Do we have to move off the tools we already use?
No, and ProductOS is not a one for one replacement for your IDE, your cloud, your database, your analytics, your SEO tooling or your automation stack. It syncs code to your GitHub, deploys to your hosting, and leaves the rest of your stack where it is.
What does it cost to run a pod?
The Team workspace is $599 per month at the current founding price and includes 10 active products, 5 User Licenses, 10 Collaborator Licenses, 15,000 monthly AI credits and 8 concurrent AI agents. Full allowances, add-ons and the 7-day free trial terms are on the pricing page.
Bring one signal.
Start with a single customer signal and a single iteration. Keep the parts that earn their place, and keep the code either way.
The sprint request goes to our contact form. Tell us the customer signal you want to act on and we scope the pod with you.