Stop babysitting your agents... — Brandon Walsenuk, Unblocked

By ai.engineer

Categories: AI, Tools

Summary

The real bottleneck with AI agents isn't intelligence—it's context. Brandon Walsenuk reveals how to stop babysitting agents by building a "context engine" that pulls static documentation AND runtime data across your entire tech stack, enabling agents to make 10x better decisions without constant human oversight.

Key Takeaways

  1. Agents spawn with zero organizational context, like a brilliant engineer on day one. You must build context systematically through documentation (MD files, agent guides) plus runtime data access across all SaaS apps and systems to avoid agents reinventing solutions that already exist.
  2. Access ≠ understanding. Tools like MCPs give agents pipes to data sources, but agents don't know what they don't know—just like new engineers. Build curated context layers that exhaustively surface patterns, existing services, and organizational conventions before agents act.
  3. A proper context engine combines three layers: static curated content (docs, patterns), runtime data queries (across multiple systems), and token-optimized summaries delivered to agents. This front-loaded context makes every downstream agent action more accurate and efficient.
  4. Follow the onboarding ladder: solo context → curated static repos → dynamic context engine. Most teams are stuck at layer one (humans as context engines). Moving to automated context retrieval unlocks autonomous agent execution at scale.
  5. Unblocked open-sourced a repo this week containing one reusable context engine component builders can extract and integrate into their own organizations without building from scratch.

Related topics

Transcript Excerpt

Hello everybody. It's good to see you. I'm Brandon. I work at Unblocked. It's a great place. Uh and my goal is to make it so that you don't have to babysit your agents anymore. Um I'm sure we all have a different take on what that means. What I think of is care and feeding. Basically agents whenever you spawn it by typing claude in your CLI let's assume whatever tool you may use they exist and it's like a brilliant software engineer has just spawned and it knows nothing about what it needs to do. It knows nothing about your org. It's completely zero context in its head. So typically what happens is people have to move through building that context which we'll go through in the beginning. But first, what I'm going to cover today pretty quickly is three myths about how you can stop babysitti…

More from ai.engineer