Editor's Notes

This week's videos highlight three concurrent trends: the shift from static AI models to 'living systems' that learn from user feedback (Trajectory.ai), the rise of AI agents as personal productivity tools deeply integrated with knowledge bases (Hermes Agent + Obsidian), and the rapid professionalization of the AI engineering field as evidenced by the expanded AIE World's Fair 2026. Together, these stories illustrate how AI is moving from research labs into continuously adapting, user-facing applications and a maturing industry ecosystem.

Key Takeaways

1. Trajectory.ai's open-source concurrent training stack lowers the barrier for companies implementing continual learning on user interactions — expect more domain-specific models that improve post-deployment.
2. Combine AI agents with local knowledge stores like Obsidian to create a persistent second brain; the agent can automate tasks while you visually manage your notes.
3. AIE World's Fair 2026 introduces specialized tracks (e.g., post-training, memory, inference) and a 'token billionaire' program — signalling that AI engineering is fragmenting into deep sub-disciplines.
4. Hermes agent's VPS-based sync enables multi-devance AI workflows; this pattern of 'agent + persistent storage' is becoming a standard architecture for personal AI assistants.
5. Companies like Harvey using Trajectory.ai show that continual learning is already viable in high-stakes domains like legal — expect more regulated industries to follow.
6. The AIE event's new engineer orientation and kids' event indicate a conscious effort to scale the AI workforce from entry-level to leadership.