Build the whole app.Built on trust.
From idea to production — architecture, identity, web, mobile and backend working as one system. Your code stays yours.
Quick check before preview
Idea → Blueprint → Apps
Lines resolve. Nodes complete. Identity stays with the request.
Modern software is too fragmented.
Frontend, backend, auth, data, payments and deploy are usually five vendors and a pile of glue. AI on top of that ships the mess faster.
Assembled, not designed
Glue everything together. Hope nothing breaks.
Pubflow starts from one plan.
Assembled, not designed
Idea → AI → Code → 10 services → Glue → Debt
Pubflow
Idea → Blueprint → Apps → Trust → Deploy
One plan. Identity in the foundation.
One system. Every application.
Web, mobile and backend share a Blueprint and a Trust Layer — instead of hoping each piece remembers the others.
Describe the product
Start with the idea in the composer. The agent works from the product, not a single file.
Keep a Blueprint
The shared plan: what it is, which apps exist, what data lives there, and who can enter. Architecture you can inspect.
Flowless is the default identity
The trust core — who someone is and what they can do — moves with the request. Not leftover middleware.
Identity moves with the request. Use it managed on Pubflow · or take the protocol.
Preview frontend and backend together
See the client and the API in one place. Download the source. Keep Cursor or Claude Code if you want.
Speed
AI understands the whole product — apps, data, and APIs — not one file at a time.
Trust
Identity and authorization sit in the foundation, not as leftover middleware.
Freedom
Build here, or keep Cursor and Claude Code. Download the source. Host it where you decide.
Assembling the plan…
Give the agent a plan it can keep.
Blueprint turns product intent into inspectable architecture so long-running work does not depend on a disappearing chat summary.
- Product features
- App responsibilities
- Data model
- API and UI plans
- Security decisions
- Risks and build order
Speed shouldn't come at the cost of trust.
Pubflow connects identity, sessions and authorization to the applications that use them — so security is architecture, not leftover middleware.
See the trust coreThe trust core
Flowless is the open-source trust core. On Pubflow, managed Flowless is the default hosted identity.
Protocol, schemas and reference implementations live at flowless.dev
Use the stack you want.
Pubflow can build the whole app on the platform. You can also use only managed Flowless and keep writing in Cursor, Claude Code, or another AI. Your stack doesn't have to adapt to Pubflow.
Build with Pubflow. Own what you build.
Generated source, inspectable architecture, downloadable applications, self-hostable runtimes.
Portable source
Download and self-host the generated application code on your preferred infrastructure.
Visible architecture
Review the Blueprint, agent plan, tools, files and previews instead of trusting a black box.
Production-oriented foundation
Core components start from patterns used in production; validation, configuration and final deployment remain explicit steps.
Other tools think in files. Pubflow thinks in systems.
Project Assistant coordinates the whole product. Coding Agent works deeply inside one application.
Know what you are building with.
A clear beta experience without hiding the current boundaries.
Does Pubflow only generate interfaces?
No. Pubflow supports complete projects with Node.js or Python backends and React, Next.js or Expo clients grouped under one Blueprint.
Can I keep and host the generated code?
Yes. You can download, modify and self-host generated application code. Managed dependencies are inventoried and documented; replacing one may require engineering, so Pubflow does not promise zero dependency.
What is the difference between Project Assistant and Coding Agent?
Project Assistant coordinates multiple Active Apps and their shared product context. Coding Agent works deeply inside one application's files, tools and preview.
Do I have to use managed identity?
Flowless is the open-source trust core for identity, sessions, and authorization. On Pubflow, managed Flowless is the default hosted identity — not an optional add-on. You can still export the source and self-host.
Which databases does Pubflow support?
PostgreSQL and compatible providers include PostgreSQL, CockroachDB, Neon and Supabase Postgres. MySQL and PlanetScale cover the MySQL family. Turso and LibSQL cover the LibSQL family. Other databases can connect through a runtime adapter; exact capabilities depend on that adapter and provider.
Can I bring my own database?
Yes. Connect an existing compatible database, or ask Pubflow to provision a managed Cloud Database. Choosing a database provider does not change the trust contract between your applications.
Which languages and frameworks work with Pubflow?
The Trust Layer works with any language or framework that can consume HTTP. The AI Builder currently generates Node.js/Python backends and React, Next.js and Expo clients. Official libraries or starters also exist for Node, Python, Go, Rust, Elixir, React and React Native, with maturity documented individually.
Can I keep using Cursor or Claude Code?
Yes. Pubflow can build the whole app on the platform. You can also use only managed Flowless and keep building in Cursor, Claude Code, or another AI against your own repo.
Your idea deserves the whole application.
Start with a private idea, make the architecture visible, and build every app with context you control.
Create with PubflowPublic beta · Build, preview, export and self-host today