# Robert Overweg - One Brain, No Filtering - AI Native DevCon June 2026

## Executive summary

The presentation details a 'One Brain' concept—a centralized, AI-native knowledge management layer designed to eliminate information silos and improve decision-making by weaving together research, client context, and operational data. The system uses an orchestrator (OpenClaw) and structured vaults to allow agents to access and synthesize organizational knowledge in real time, shifting focus from manual file retrieval to idea generation and proactive insights.

## Key takeaways

- Shift from File Search to Idea Synthesis: The core value lies in moving beyond searching for specific files; the system allows users to search for 'ideas' or 'contacts.' Agents can interpret natural language queries (e.g., asking about CI/CD steps) and provide contextually accurate answers based on stored knowledge.
- Structured Knowledge Flow: Knowledge is categorized into 'company knowledge' (new developments, research wikis) and the 'creation pipeline.' Information must be promoted to a central vault from various sources (e.g., Obsidian notes, meeting transcripts) to gain grounding in reality before being shared widely.
- Scaling and Security Challenges: While the system is powerful, scaling remains a challenge, particularly regarding data segregation (permissions) across different client or team buckets. The local setup on one person's laptop was initially used for testing, but enterprise rollout requires careful consideration of security boundaries.

## Technical details

- Architecture and Orchestration: The system uses OpenClaw as the orchestrator, managing structured vaults. The architecture is designed to interpret diverse inputs (Miro maps, PowerPoint, Figma) and normalize them into a single knowledge base.
- Knowledge Storage and Indexing: Data sources include GitHub repositories (for research), Obsidian (local browsing/editing), and structured vaults. Advanced indexing utilizes vector keywords, graph relations, and semantic search over memory files.
- Automation and Agents: AI agents are used for proactive tasks, such as tracking accounts on X (Twitter) via cron jobs or analyzing external skills (e.g., Addy Osmani's skill breakdown) against the existing codebase to determine relevance.
- Meeting and Content Capture: Tools like Omi (open-source recording), Granola, and automated transcription services are used to capture all conversations. These transcripts feed into the system, allowing for semi-automatic generation of first drafts of presentations or documents.

## Practical implications

- Implementing a centralized knowledge vault drastically reduces time spent on manual information retrieval and context switching.
- The system enables the creation of 'Chief of Staff' agents that surface necessary questions or tasks during meetings, optimizing human effort.
- By making all conversations (meetings, research) available to the AI, organizations can reduce reliance on traditional hierarchical coordination mechanisms.

## Topics

AI Agents, Knowledge Management Systems (KMS), DevOps Automation, Semantic Search, Information Architecture, OpenClaw, Obsidian, GitHub, Neo4j, Codex

Source: https://www.youtube.com/watch?v=rmxRlpi7xN4