OpenClaw + AlphaClaw + GBrain

Why deploy openclaw gbrain on Render?

OpenClaw GBrain is a one-click deployment template that bundles the OpenClaw AI agent platform with GBrain, a Postgres-native knowledge store providing hybrid search (vector + keyword + ranking fusion). It solves the cold-start problem for AI agents by giving them a persistent, searchable memory brain from first boot, running entirely in-process via embedded PGLite with no external database required.

This template wires together AlphaClaw, OpenClaw, and GBrain with an embedded PGLite database running in-process on a single container with persistent disk—no external Postgres to provision or connect. You skip configuring the hybrid search stack (vector + keyword + RRF fusion), seeding the skill pack into the right directory, and sizing memory for in-process database workloads. Render's persistent disk keeps your brain file and agent state intact across deploys, and the Pro plan default gives you headroom for bulk ingestion without manual scaling decisions.

Architecture

What you can build

After deploying, you'll have an OpenClaw AI agent running with a persistent knowledge base that supports hybrid search across anything you feed it. You can immediately start importing markdown files and querying your knowledge through natural conversation—the agent already knows how to ingest, search, and summarize without any additional configuration. Everything runs in a single container with an embedded database, so there's no external infrastructure to manage.

Key features

• PGLite embedded Postgres: Runs Postgres compiled to WebAssembly in-process with pgvector and pg_trgm, eliminating external database dependencies.
• Hybrid search with RRF fusion: Combines vector similarity, keyword matching, and reciprocal rank fusion with multi-query expansion for knowledge retrieval.
• Pre-seeded skill pack: GBrain skills for ingest, query, maintain, enrich, and briefing are auto-discovered by OpenClaw on first boot.
• Single container architecture: AlphaClaw, OpenClaw, and the knowledge brain run in one container with one persistent disk, avoiding multi-service billing.
• Idempotent schema migrations: The gbrain init command automatically applies pending migrations on boot, with manual re-run available via SSH for major upgrades.

Use cases

• Solo founder gives AI agent searchable memory of all company docs
• Developer imports markdown knowledge base for instant hybrid search queries
• Startup deploys persistent AI brain without managing external Postgres database
• Technical writer asks agent for briefings from ingested documentation

What's included

Prerequisites

• OpenAI API Key: Required for GBrain to generate text embeddings using the text-embedding-3-large model for vector search.
• Anthropic API Key: Required for GBrain's multi-query expansion and LLM-powered chunking features using Claude Haiku.

Next steps

1. Open your Render service URL and complete the AlphaClaw welcome wizard — You should see the setup screen prompting for your SETUP_PASSWORD, then a configuration wizard that finishes with a ready chat interface
2. Test the GBrain connection by asking 'How many pages are in the brain right now?' — The agent should respond with a count of 0 pages, confirming GBrain is initialized and the skill pack is working
3. Upload a markdown file by pasting its contents into the chat and saying 'Add this to the brain' — You should see confirmation that the document was ingested, and a follow-up search for a term from that document should return it in results

Resources

• OpenClaw official site