Smarter Apps with Ash and GenAI
summary
Ash resources become a machine-readable API that an LLM can call: a ~500-line Ash mCP server exposes actions as tools (reads, search, create) via OpenAI function calling and the new mCP protocol. Live demo returns exactly three artists, denies create when policy requires admin then succeeds after admin role, and even infers country and album links from cover URLs—unexpectedly smart. Limits include 64-tool suggestion cap, stdio-only mCP today, small context windows and production safety concerns.
AI-written summary · about
chapters (12)
- 0:00Introduction to the speaker
- 1:15The problem Ash solves
- 4:08Understanding Ash fundamentals
- 8:27LLMs in the future of software
- 12:07Tunez app architecture
- 14:19Tool usage and Ash AI
- 22:23Model Context Protocol
- 26:01Implementing MCP in Ash
- 27:21Future challenges and goals
- 31:34Resources and training
- 32:50Q&A session
- 35:57CodeBEAM conference details
related talks
01h 28mAI Coding Agents and how to code them
54mFrom Single-Shot LLMs to Intelligent Agents: Building Scalable AI Systems with Spring AI and MCP
45mAgents, Tools, and MCP, oh my! Next-level AI concepts for developers
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
23mMCP Mashups: How AI Agents are Reviving the Programmable Web
47mOutsmarting the Genie: Smart Architecture Decisions for GenAI Apps
01h 01mAI Agents In-Depth – Function Calling, MCP and Tool Use Under the Hood
01h 03mAI Coding Agents and how to code them