What is Forginate?
Forginate is an AI software factory. You describe the software you want, and Forginate takes it through a complete delivery pipeline — from intake and discovery, through planning documents you review and approve, to an AI build phase, automated review and QA, and finally deployment to infrastructure you control or infrastructure Forginate provisions for you.
Think of it as hiring a disciplined engineering team that writes everything down first: before any code is written, Forginate produces a structured set of planning documents, asks you to approve them, and only then starts building. After delivery, the product stays alive — you file improvement requests, and Forginate plans, builds, and ships each one as a reviewed change.
Forginate is hosted at dev.forginate.com.
What you can do with it
- Build new software from scratch. File a "New project" request, answer discovery questions, approve the plan, and watch the build run.
- Import an existing codebase. The "Import existing" request type reverse-engineers planning documents from a repository so you can iterate on it through the same pipeline.
- Adopt a production app. "Import production" brings an already-live application into maintenance mode.
- Evolve shipped products. Features, changes, and bug fixes are filed as improvement requests on a project's Improve board, where they are triaged, planned, approved, built, QA'd, and merged.
- Deploy anywhere. Ship to your own Kubernetes cluster via a lightweight deploy agent, or let Forginate provision a dedicated AWS account and run your app on ECS Fargate.
- Collaborate. Invite teammates with per-project roles, and share read-only document portals with external stakeholders — no account required on their side.
How it's organized
Everything in Forginate belongs to an organization:
- An organization owns projects — each project is one product moving through the factory.
- A project accumulates requests (units of work), documents (the versioned plan for the product), runs (individual AI agent executions), and deployments.
- Organization members have roles, and access to individual projects is granted separately, so you can keep client work isolated within one org.
See Core concepts for the full model.
The pipeline at a glance
The Pipeline board in the app shows every request moving through these phases:
| Phase | What happens |
|---|---|
| Intake | Your request is submitted and classified |
| Define | The document plan is drawn up (or, for imports, reverse-engineered) |
| Plan | Planning documents are generated section by section for your review |
| Build | Build agents write the code, task by task |
| Integrate | The work is assembled, audited, and gaps are closed |
| Deliver | Final approval, then delivery |
| Improving | The product is live and open to improvement requests |
Every phase that produces documents puts them in front of you with an approval gate. Nothing advances past a gate without your sign-off.
Humans stay in the loop
Forginate's AI does the writing and building, but you approve the table of contents, the planning documents, the deployment plan, and the final delivery. You can edit any document directly or through the conversational vibe editor before approving it.
What Forginate is not
- It is not a code autocomplete tool or an IDE plugin. It runs the whole delivery process.
- It is not a black box. Every agent run, its cost, and its output are visible in the UI, and every document is versioned with a full history.
- It does not require you to hand over production credentials. The Kubernetes deploy agent connects outbound to Forginate — you never open inbound firewall ports — and AWS deployments run in a dedicated account provisioned for you.
Cost transparency
AI work costs money, and Forginate shows you exactly where it goes:
- Every run displays its token usage and cost in the UI.
- Planning shows a cost estimate before you approve the table of contents.
- Runs have budget caps you can raise mid-run if you want the work to continue.
- You can use Forginate's platform-managed models (metered billing) or bring your own Anthropic, OpenAI, or compatible API keys with cost passthrough. See Billing.
Where to go next
- Getting started — create your first project in minutes.
- Pipeline overview — a deeper walk through each phase.
- Deployments — how shipping to real infrastructure works.