DevOps and Docker Talk: Cloud Native Interviews and Tooling Heroku + MCP = The Fastest Way to Run AI Agents in the Cloud
9 snips
Jun 6, 2025 Julián Duque, Principal Developer Advocate at Heroku, dives into the transformative capabilities of their AI platform. He discusses Heroku's shift to Kubernetes and the significance of Master Control Points (MCP), which streamline workflows and boost productivity. Listeners get insights into deploying AI-driven applications seamlessly using new features like PG Vector and integrating vector databases. Duque also highlights future prospects for Heroku's AI tools, empowering developers to leverage AI more effectively in their projects.
AI Snips
Chapters
Transcript
Episode notes
Heroku's Managed AI Inference
- Heroku offers managed AI inference where models run securely on their infrastructure, preserving private data privacy.
- It supports several LLMs and allows local tool execution with languages like Python, Ruby, Go, and Node.js.
Deploy MCPs Easily on Heroku
- Deploy local MCPs to Heroku with a single command to integrate seamlessly with inference services.
- Use the MCP gateway with authentication to expose those MCPs via HTTP endpoints.
Using Figma MCP for CSS
- Julián used the Figma MCP to convert design files into Tailwind CSS automatically.
- This significantly sped up his frontend development despite not being a designer.