Deploy your own Launchstack
in under an hour.
Launchstack is the open-source second brain for founders — docs, recordings, and messages turned into a cited knowledge graph. Bring your own API keys, host it anywhere, keep your data.
What you're shipping
Everything in the box when you spin up a fresh Launchstack instance.
RAG over everything
PDFs, Office docs, transcripts, repos, and exports — chunked, embedded, and retrieved with pgvector plus optional reranking.
Predictive analysis
Eight document classes (contract, financial, compliance, technical, HR, research, educational, general) with per-type recommendations.
Legal generation
Template-driven DOCX for NDAs, employment, and service agreements, with clause-level AI refinement.
Voice & transcription
Gemini turns audio and video uploads into searchable documents, and speaks answers back — one key for both.
Marketing pipeline
Generate and verify platform-specific posts for Reddit, X, LinkedIn, and Bluesky against your own source docs.
Pluggable storage
Vercel Blob by default. S3, UploadThing, and local disk swap in through the storage port.
Get started
Four accounts to create before your first boot.
Clerk
An AI provider
PostgreSQL with pgvector
postgres + pgvector container for self-hosting.Vercel Blob
BLOB_READ_WRITE_TOKEN automatically.Quick start
Run the app on your machine in five commands.
Clone the repository
git clone https://github.com/Deodat-Lawson/pdr_ai_v2.git
cd pdr_ai_v2Install dependencies
pnpm installConfigure environment variables
Drop a .env at the repo root with at least these keys.
DATABASE_URL="postgresql://user:password@host:5432/db?sslmode=require"
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=<your-clerk-publishable-key>
CLERK_SECRET_KEY=<your-clerk-secret-key>
# One OpenAI-compatible endpoint — Gemini shown, see Chat Models for others.
# Model ids and route assignments live in apps/web/config/chat-models.yaml.
CHAT_BASE_URL=https://generativelanguage.googleapis.com/v1beta/openai
CHAT_API_KEY=<your-google-ai-key>
# Vercel Blob — required for document uploads
BLOB_READ_WRITE_TOKEN=vercel_blob_rw_xxxxxxxxxxxx
# Inngest — placeholder is fine for local dev
INNGEST_EVENT_KEY=dev-placeholderApply the schema migrations
pnpm --filter @launchstack/core db:migrateStart the dev server
pnpm --filter @launchstack/web devChoose a deployment path
Pick what matches your infra.
Vercel
Managed hosting with auto-deploys from GitHub. Neon serverless Postgres handles pgvector out of the box.
Docker
Self-host the full stack with Docker Compose: Postgres + pgvector, migrate, app, and optional OCR sidecars.
Required integrations
Must be configured before the app boots cleanly.
AI providers
Google Gemini out of the box. Any endpoint speaking the OpenAI chat-completions protocol also works, including a local vLLM or Ollama server.
Vercel Blob
Primary document storage backend. There is no database-only fallback for uploaded files.
Inngest
Event-driven background jobs for OCR, embeddings, analysis, and scheduled content pipelines.
Optional integrations
Enable as needed.
LangSmith tracing
Observability for every LLM, retriever, and tool call.
OCR services
Azure Document Intelligence or Landing.AI for scanned and complex layouts.
Exa search
Semantic web search for the trend and research agents.
Voice & audio
Gemini voices for spoken answers. Transcription is built in on the same key.
Video walkthroughs
Short clips for the trickier steps.
.env.example. Put real keys in your deploy platform’s environment variable manager.