I Gave an AI Agent the Keys to My Life (Here's What Happened)
summary
Claims stepwise permission creep: OpenClaw gains access to email, files, calendar and a 3,000 Obsidian pages vault, builds search and memory, runs scripted jobs and LLM judgment, and performs overnight maintenance and backups (cron jobs at 2am). Concrete results: auto-renewed a domain, fixed a Netflix payment failure, drafted business replies into a drafts folder, and rebuilt indexes after failures. Includes overnight cron ecosystem and a tuned agent personality that sometimes disagrees.
AI-written summary · about
chapters (10)
- 0:00Start
- 0:15Radek's path to OpenClaw
- 2:17The philosophy of incremental growth and system updates
- 4:51Integrating the Obsidian knowledge base
- 8:59Ambient operations and overnight automation
- 11:02Core job types for the AI agent (Ambient Operations, Attention Filtering, Execution)
- 13:03Deep dive into specific Discord integration channels
- 14:54System architecture: LLMs, scripts, and memory management
- 16:28Challenges: Bad memory, brittle automations, and noisy nodes
- 17:19Conclusion: Optimizing for the future self
related talks
46mAI Agents Are Becoming Operating Systems: Lessons From OpenClaw
30mAI Agents Unchained
24mAI Agent Swarms Are Amazing
17mDon't give your agent the master key
30mManaging your life with AI
31mOops, My AI Agent Just Deleted All My Email: Locking Down Agents With Signed Policies
49mAI Agents: Collaboration in Real-Life Applications
11mBuilding Secure Backend Services with AI Agents