AI-native SaaS foundation
An AI SaaS boilerplate for the product and the developer
Boilerplate-Stack supports AI on both sides of the repository. Your customers get multi-provider chat, agents, RAG and usage accounting. Your coding agent gets the rules, skills and architecture it needs to extend those features without inventing a second system.
Ship AI features on an existing product boundary
- OpenAI, Anthropic and Google models through direct APIs or OpenRouter
- Streaming chat, built-in agents and document-grounded RAG
- Account-scoped credits, token usage and cost analytics
- Repository rules for Claude Code, Codex, Cursor and AGENTS.md tools
AI behavior remains observable and controlled
One server-only provider boundary
Chat and embedding requests cross one LLM client that owns model selection, routing, timeout, retry and fallback behavior. Provider SDKs do not spread through routes and components, so policy changes have one implementation point.
Usage and product access share an account
AI requests, chat sessions, documents and credit transactions resolve through the same personal or workspace account. The stream checks available access before generation and records the provider usage needed for accounting afterward.
The repository teaches the coding agent
Domain rules describe the project’s API, database, billing, security and localization conventions. Repeatable skills and specialist reviewers give AI coding tools concrete workflows instead of relying on one enormous prompt.
AI capabilities already implemented
- Streaming chat with persisted user and assistant history
- RAG ingestion for PDF, Markdown and text documents
- Cross-provider fallback, circuit breaker and bounded retries
- AI transparency, durable output marking and provider disclosure
Give both your users and your coding agent a real foundation
Start with the provider boundary, accounting and project context already in place, then build the AI workflow only your product needs.
See pricing