Building Agentic Applications w/ Heroku Managed Inference and Agents
summary
Heroku provisions LLMs (Claude 3.5/3.7/4, Cohere embeddings, Stable Image Ultra) as a platform add‑on and runs an agentic control loop that calls tools on one‑off dynos which scale to zero. Native tools include dyno run command, postgres‑get‑schema and postgres‑run‑query, plus code execution in Python/Node/Ruby/Go with on‑the‑fly dependency installs. Demos show streaming chat, chained tools (web scrape → run code), and an example that computes the 30th Fibonacci number (demo returns 233).
AI-written summary · about
chapters (16)
- 0:00Welcome and setup
- 3:35Heroku AI overview
- 8:12Workshop prerequisites
- 13:31Deploying Jupyter notebook
- 16:22Provisioning managed inference
- 17:06Basic inference examples
- 21:02Streaming inference demo
- 24:56Introducing agents
- 28:33Dyno command execution
- 31:02Code execution tools
- 33:33Chaining tools together
- 36:03Database access tools
- 43:14Deploying MCP servers
- 46:29Using external MCPs
- 48:17Remote MCP access
- 51:05OpenAI SDK compatibility
related talks
28mBuilding Agentic AI applications on AWS
32mStop Chat-and-Hope and Start Building Better Apps with Agents
21mServerless Agents: Real-World Tooling with Strands SDK, MCP & AWS
43mThe AgentVerse - Build fast, Scale smarter
16mBuilding Agentic AI applications with Positron and AWS Strands Agents
11mRuby startup demo – Arjun Singh
15mServerless & Agentic AI: Better Together
30mBuilding Robust, Secure LLM and Agentic AI Workflows