Building AI Workflows: From Local Experiments To Serving Users
summary
Claims a repeatable path from local LLM hacks to deployable apps using containerized, static configs and shows concrete tooling: Docker model runner for local OpenAI‑compatible endpoints, an MCP toolkit/gateway for tool catalogs and secret isolation, Compose files to package agents, and examples using goose by block plus TTYD and Cloudflare tunneling. Demonstrates a docker‑compose agent that runs a 3B quantized model, wires to a YouTube‑transcript MCP, exposes a browser terminal, triggers GPU load and serves remotely, proving the approach works end to end.
AI-written summary · about
chapters (9)
- 0:00AI workflows as applications
- 4:57AI agents as applications
- 7:25LLMs locally with Docker
- 10:35Connecting tools via MCP
- 13:23Packaging agents with Docker Compose
- 18:23Deploying agents as apps
- 21:14Systems of agents with Compose
- 23:20Cloud providers and containerized AI
- 24:02Closing thoughts on repeatable AI
related talks
30mMind the Gap - A Developer's Roadmap To Building AI Agents
39mBuilding Agentic AI Apps with Docker MCP & Cagent
50mCompose the Future: Building Agentic Applications, Made Simple
46mDocker Compose Your Dev Toolkit for AI & Cloud
37mUniversal AI: Execute Your Models Where Your Data (And Users) Are
43mDocker Agent — How to Make Building AI Agents Even Simpler?
07mKeynote: Accelerating GenAI Adoption with Containers: Unleashing AI/ML Development...
01h 42mBuilding Responsible AI Agents in Decentralized Open-Source Environments