Phase 15

Autonomous Systems

Phase 15: Autonomous Systems. 22 hands-on lessons building AI from first principles in the browser. Free reading; graded exercises and certificate with lifetime access.

  1. The Shift from Chatbots to Long-Horizon Agents (graded)
  2. STaR, V-STaR, Quiet-STaR — Self-Taught Reasoning
  3. AlphaEvolve — Evolutionary Coding Agents
  4. Darwin Godel Machine — Open-Ended Self-Modifying Agents
  5. AI Scientist v2 — Workshop-Level Autonomous Research
  6. Automated Alignment Research (Anthropic AAR)
  7. Recursive Self-Improvement — Capability vs Alignment (graded)
  8. Bounded Self-Improvement Designs
  9. The Autonomous Coding Agent Landscape (2026)
  10. Claude Code as an Autonomous Agent: Permission Modes and Auto Mode
  11. Browser Agents and Long-Horizon Web Tasks
  12. Long-Running Background Agents: Durable Execution
  13. Action Budgets, Iteration Caps, and Cost Governors (graded)
  14. Kill Switches, Circuit Breakers, and Canary Tokens (graded)
  15. Human-in-the-Loop: Propose-Then-Commit
  16. Checkpoints and Rollback
  17. Constitutional AI and Rule Overrides
  18. Llama Guard and Input/Output Classification
  19. Anthropic Responsible Scaling Policy v3.0
  20. OpenAI Preparedness Framework and DeepMind Frontier Safety Framework
  21. METR Time Horizons and External Capability Evaluation
  22. CAIS, CAISI, and Societal-Scale Risk
0 lifetime access. Curriculum based on AI Engineering from Scratch by Rohit Ghumare (MIT, used under attribution).