Design Partner Program: Early Access to Agent-Native Knowledge Bases
We are looking for 10 design partners to shape the next generation of AI agent memory.
The Problem
AI agents forget everything between sessions. Your agent debugged a production incident yesterday and has no memory of it today. The architecture decisions documented last month vanish when the context window closes.
Vector search helps — but only for questions where "similar-sounding documents" equals "relevant context." Enterprise knowledge has structure: modules depend on modules, configurations cascade across systems, workflows connect across departments. Cosine similarity does not capture these relationships.
What We Built
The Agent-Native Knowledge Base on agent.ceo combines a Neo4j knowledge graph with HNSW vector search. Agents read from it, write to it, and traverse organizational knowledge programmatically.
- Graph + Vector: Neo4j for typed relationships between entities. HNSW vector index for semantic search. Agents use both — vector search finds the starting point, graph traversal finds the context.
- 26 MCP Tools: Full programmatic access — search, create, update, traverse, ingest. Agents interact with the knowledge base as a first-class tool, not a separate system.
- PKCE OAuth 2.0: Secure authentication for external tools. Claude Code, CI pipelines, and agent pods authenticate without API keys or shared secrets.
- Multi-Tenant Isolation: Per-organization knowledge spaces with granular permissions. Per-page access control. No cross-org data leakage.
We recently ingested 5,000+ pages of enterprise ERP documentation into a single knowledge graph. Agents answer cross-module dependency questions that no search engine can.
What Design Partners Get
Priority access to the knowledge base platform, including features still in development.
Direct founder access. Weekly calls, shared Slack channel, same-day response on issues. You are not talking to support — you are talking to the person who built it.
Roadmap influence. Your use cases shape what we build next. Design partner feedback has already driven three major features: wikilink traversal, freshness scoring, and bulk ingestion from cloud storage.
Dedicated onboarding. We help you ingest your documentation, configure agent access, and validate the knowledge graph structure. Not a self-serve tutorial — hands-on engineering support.
Who We Are Looking For
Engineering teams that meet at least two of these criteria:
- Running AI agents in production (or actively building toward production deployment)
- 5,000+ pages of internal documentation (Confluence, SharePoint, custom wikis, or exported docs)
- Multi-agent or multi-team AI usage where shared memory across agents or sessions is a bottleneck
- Enterprise software domain with complex interdependencies between systems, modules, or configurations
Verticals where we see the strongest fit: enterprise software vendors, system integrators, large SaaS platforms, and regulated industries where audit trails matter.
What We Ask
- A 15-minute intro call (live demo, no slides)
- Honest feedback on what works and what does not
- Permission to reference your use case anonymously in our documentation (no company names without explicit agreement)
Get Started
Email moshe@genbrain.ai with a one-paragraph description of your documentation challenge. We will respond within 24 hours with a demo link.
Related
- Introducing Agent-Native Knowledge Bases — the KB launch announcement
- Enterprise Knowledge Ingestion: 5,000 ERP Pages in One Command — what the ingestion pipeline does
- Multi-Tenant Agent Orchestration — how we isolate organizations and agents