The AI Engineering Operating System
Stop prompting AI. Start engineering with AI.
Build production-grade software through Specification-Driven AI Engineering (SDAE).
AI coding tools have dramatically reduced the cost of writing code.
But software development is more than writing code.
The difficult questions still remain:
- What should we build?
- Why should we build it?
- What should we not build?
- What architecture should we choose?
- How should the business domain be modeled?
- How should APIs evolve?
- How do we build securely?
- How do we break work into small, reviewable changes?
- How do we validate AI-generated software?
FounderOS addresses the engineering process around AI β not just code generation.
Human
β
βΌ
Prompt
β
βΌ
AI
β
βΌ
Code
Product Intent
β
βΌ
Product Discovery
β
βΌ
Specification
β
βΌ
Architecture
β
βΌ
Implementation Plan
β
βΌ
AI Engineering Agents
β
βΌ
Incremental Implementation
β
βΌ
Validation
β
βΌ
Production Software
FounderOS enables founders and small engineering teams to build software with the discipline of a world-class engineering organization.
Instead of treating AI as a code generator, FounderOS treats AI as a team of specialized engineering roles working through structured workflows and reusable engineering artifacts.
FounderOS provides:
- π€ AI Engineering Agents
- π Engineering Artifacts
- π Architecture Standards
- π Implementation Planning
- β Quality Gates
- π Explicit Handoffs
- π Reusable Templates and Playbooks
FounderOS is for founders and small engineering teams who want to use AI while keeping product decisions, specifications, and quality gates explicit.
Use it as a working method for a project:
- Start with the Quick Start and copy the relevant templates into your project workspace.
- Follow the workflow from discovery and requirements through architecture, planning, implementation, testing, and release.
- Use the Agent Catalog to choose an agent guide for each handoff, and the playbooks for repeatable procedures.
- Review and approve product, security, and release decisions as a person on the team; keep assumptions and open questions visible.
FounderOS currently provides documentation, templates, agent instructions, and playbooks. The long-term vision includes possible CLI tooling, but no CLI or npm package is available today. See the completion roadmap for current progress.
Great software is the result of great engineering systems β not better prompts.
AI makes software development faster.
FounderOS provides the engineering system that tells AI how to work.
Think
β
Discover
β
Specify
β
Design
β
Plan
β
Build
β
Validate
β
Ship
β
Learn
FounderOS organizes AI capabilities around engineering roles, not AI vendors.
The framework is intentionally tool-agnostic and can work alongside Kiro, Cursor, Claude Code, ChatGPT, Gemini, or future AI development tools.
| Engineering Role | Agent | Status |
|---|---|---|
| π¨βπΌ Senior Product Manager | Product Discovery Agent | β |
| π Product Manager | PRD Agent | β |
| Engineering Role | Agent | Status |
|---|---|---|
| π Staff Engineer | Solution Architecture Agent | β |
| π Database Architect | Database Design Agent | β |
| π§© Domain Architect | Domain Modeling Agent | β |
| π API Architect | API Specification Agent | β |
| π Principal Security Engineer | Security Architecture Agent | β |
| Engineering Role | Agent | Status |
|---|---|---|
| π Staff Engineer / Engineering Manager | Implementation Planning Agent | β |
| π» Senior Backend Engineer | Backend Module Implementation Agent | β |
| π¨ Senior Frontend Engineer | Frontend Module Implementation Agent | β |
| π€ AI Platform Engineer | AI Module Implementation Agent | β |
| π§ͺ QA Lead | Testing Agent | β |
| π Principal Engineer | Code Review Agent | β |
| π DevOps Engineer | Release Agent | β |
FounderOS agents collaborate through structured engineering artifacts.
Agent
β
βΌ
Engineering Artifact
β
βΌ
Next Agent
Examples:
| Agent | Artifact |
|---|---|
| Product Discovery | Product Discovery Document |
| PRD | Product Requirements Document |
| Solution Architecture | Architecture Document |
| Database Design | Database Design Document |
| Domain Modeling | Domain Model |
| API Specification | API / OpenAPI Specification |
| Security Architecture | Security Architecture |
| Implementation Planning | Implementation Plan |
| Backend Implementation | Backend Module |
| Frontend Implementation | Frontend Module |
This creates a traceable engineering workflow instead of a collection of disconnected prompts.
π‘
IDEA
β
βΌ
π DISCOVER
β
βΌ
π SPECIFY
β
βΌ
π DESIGN
β
βΌ
π PLAN
β
βΌ
π€ BUILD
β
βΌ
π§ͺ VALIDATE
β
βΌ
π SHIP
β
βΌ
π LEARN
β
ββββββββββββββββΊ Improve
The goal isn't to eliminate engineering judgment.
The goal is to make good engineering judgment repeatable.
FounderOS is not:
- β Another AI coding assistant
- β Another collection of random prompts
- β A replacement for Kiro, Cursor, Claude Code, or ChatGPT
- β A framework locked to a specific AI vendor
- β A "generate my entire application" workflow
FounderOS sits above these tools.
It defines how AI participates in software engineering.
FounderOS/
β
βββ README.md
β
βββ docs/
β βββ vision.md
β βββ workflow.md
β βββ principles.md
β
βββ agents/
β βββ product/
β β βββ product-discovery-agent.md
β β βββ prd-agent.md
β β
β βββ architecture/
β β βββ solution-architecture-agent.md
β β βββ database-design-agent.md
β β βββ domain-modeling-agent.md
β β βββ api-specification-agent.md
β β βββ security-architecture-agent.md
β β
β βββ engineering/
β β βββ implementation-planning-agent.md
β β βββ backend-module-implementation-agent.md
β β βββ frontend-module-implementation-agent.md
β β
β βββ frontend/
β βββ ai/
β βββ shared/
β
βββ templates/
β βββ product-discovery-document.md
β βββ prd.md
β βββ ...
βββ playbooks/ # Operational playbooks
βββ examples/
βββ assets/
FounderOS is currently in early development.
- β Vision
- β Engineering Workflow
- β Engineering Principles
- β Product Discovery Agent
- β PRD Agent
- β Solution Architecture Agent
- β Database Design Agent
- β Domain Modeling Agent
- β API Specification Agent
- β Security Architecture Agent
- β Implementation Planning Agent
- β Backend Module Implementation Agent
- β Frontend Module Implementation Agent
- β AI Module Implementation Agent
- β Testing Agent
- β Code Review Agent
- β Release Agent
The role documents exist. Canonical workflow templates and initial operating playbooks are available; the reference project is a draft design walkthrough. See the FounderOS Completion Roadmap.
Start a project with the Quick Start, or review the Agent Catalog to see current workflow ownership and gaps.
- Vision
- Workflow
- Engineering Principles
- Agent framework
Status: π’ Complete
- Product Discovery
- PRD
- Solution Architecture
- Database Design
- Domain Modeling
- API Specification
- Security Architecture
Status: π’ Complete
- Implementation Planning
- Backend Module Implementation
- Frontend Module Implementation
- AI Module Implementation
- Testing
- Code Review
- Release
Status: π’ Complete
Build canonical templates for:
- Product Discovery
- PRD
- Architecture
- Database
- Domain Model
- API
- Security
- Implementation Plan
- Testing
- ADRs
Status: π’ Complete
Create reusable playbooks for:
- Engineering
- Security
- Deployment
- Operations
- Incident Response
- AI-assisted development
Status: π‘ In Progress
Build complete products using FounderOS from:
Idea
β
Discovery
β
Specification
β
Architecture
β
Planning
β
Implementation
β
Testing
β
Release
Status: π‘ In Progress
FounderOS follows a few fundamental principles:
- Specification before implementation
- AI is part of the engineering team
- Product-first thinking
- Simplicity over unnecessary complexity
- Small, reviewable changes
- Security by design
- Documentation as a first-class artifact
- Explicit engineering decisions
- AI-assisted, human-approved
- Tools may change; engineering principles remain
See docs/principles.md.
FounderOS is not intended to remain a prompt library.
The long-term vision is an AI Engineering Operating System where specialized AI engineering agents collaborate through structured artifacts, quality gates, and explicit handoffs.
Eventually, the workflow could look like:
founderos init
founderos discover
founderos specify
founderos architect
founderos plan
founderos implement
founderos test
founderos review
founderos releaseThe CLI is a future direction. The current focus is building the methodology and proving it with real projects.
FounderOS is an early-stage experiment.
The goal is to explore a simple question:
What happens when we give AI not just coding capabilities, but the engineering process of a high-performing software organization?
The framework is evolving through real projects, experiments, and feedback.
Ideas, criticism, contributions, and experiments are welcome.
Don't give AI a prompt. Give AI an engineering system.
AI can generate code.
FounderOS provides the context, process, constraints, architecture, artifacts, and quality gates around that code.