Build AI Agents That Survive the Enterprise
Build AI agents that survive real users, messy tools, long-running workflows, security boundaries, and production constraints. Crack open the machinery behind them, from loops and harnesses to MCP, evals, guardrails, and multi-agent systems, then make every step inspectable, controllable, and trustworthy.
By continuing, you agree to Maven's Terms and Privacy Policy.
Thu Jul 23·10:00 PM UTC
Building Effective Agent Harnesses
Agent frameworks have an impressive talent for making every feature feel mandatory. Memory! Planning! Multi-agent orchestration! You ain't gonna need most of it. We’ll build an agent loop from first principles, examine MCP, hooks, cron jobs, and sub-agents, and learn from Pi, Codex, and Claude Code so you can decide what belongs in your harness and what belongs on a conference slide.
You'll learn from
Hugo Bowne-Anderson
AI & data engineer, consultant, educator of 6+ million students (ex-Yale)
Tue Jul 28·11:00 PM UTC
Loop Engineering, Without the BS
Every agent workflow now wants a loop. Generate, judge, revise, repeat, and keep spending tokens until the chart goes up. Sometimes this is useful optimization. Sometimes it is an LLM grading its own homework with extra steps. We’ll separate the engineering from the hype, decide when a loop is worth running, and build eval boundaries that stop agents from rewriting the answer key.
You'll learn from
Hugo Bowne-Anderson
AI & data engineer, consultant, educator of 6+ million students (ex-Yale)