Wednesday, September 16, 2026
5:30 PM – 8:30 PM CDT
HOTA: Beyond the Prototype II — Self-Recursive Agents
About this event
An Austin AI Alliance HOTA workshop for builders whose agent prototypes degrade in production. Instructor Mario Facussé walks through experience-learning, versioning, and human-in-the-loop reinforcement patterns with a live production stack.
The Austin AI Alliance's HOTA (Human Orchestration of Tools & Agents) series runs practitioner-focused sessions built around a single workflow, taught with enough time to actually practice it.
This edition targets a specific failure mode: agents that work as prototypes but degrade once they are running for real. Instructor Mario Facussé (Managing Director, MEF Solutions) covers how three methods fit together — Structured Experience Learning (SEL) for agents that learn from their own execution history, Autonomous Versioning Optimization (AVO) for telemetry-driven prompt evolution, and Human-Agent Collaborative Reinforcement Learning (HACRL).
Other topics include designing a Behavior Registry and Bounded Work Packets to scope agent tasks, communication discipline between agents, and the production concerns that usually go unmentioned: memory decay, pruning governance, and cost control. The session includes a live demonstration on a working stack of n8n, Supabase, Qdrant, and OpenRouter.
Held at the Center for Government and Civic Service in ACC Rio Grande, Building 3000 (1218 West Avenue). $55 for Austin AI Alliance members, $85 for non-members. Bring a laptop.
Organized by
An Austin nonprofit building an inclusive AI community through hands-on education, workshops, and events that help people learn and apply AI tools, skills, and workflows.