AWS re:Invent 2025 - From vibe to live in minutes with Heroku AI PaaS (AIM250)

AWS re:Invent 2025 - Heroku AI PaaS Summary

Introduction to Heroku AI PaaS

• Heroku is a Salesforce company and a public-traded Platform-as-a-Service (PaaS) provider
• Heroku has been in the PaaS business since 2010/2011, initially focused on Ruby on Rails applications
• Heroku now supports over 9 programming languages and various managed services
• Heroku is renowned for offering a great developer experience, leading to the creation of "Heroku of X" open-source products

Building the "Heroku of AI"

• Heroku recognized the need for a platform that provides a great developer experience for building AI applications and agents
• Building AI applications involves many complex decisions and considerations:
  • Selecting the right large language models (LLMs) for the application's needs
  • Integrating the LLMs with other applications and services
  • Securely providing the LLMs access to data
  • Operating the infrastructure behind the AI models

Heroku AI PaaS Offerings

1. Managed Inference and Agents:
   • Allows provisioning of AI models to applications with a single click or command
   • Supports a curated list of LLMs, including Cloud 4.5 Sonnet, Cloud 4.5 Haiku, Amazon Nova Light, Amazon Nova Pro, GPT-OSS 120B, and more
   • Integrates with the Model Context Protocol (MCP) to provide LLMs with access to additional context and resources
   • Includes a Heroku MCP server for developers and a platform to securely deploy custom MCP servers
2. Vector Database with PG Vector:
   • Provides a vector database solution built on top of PostgreSQL
   • Enables similarity search and the creation of retrieval-augmented generation pipelines
3. Heroku Bots:
   • A tool that allows building web applications using natural language
   • Generates a step-by-step plan for building the application based on the user's prompt
   • Automatically deploys the generated application to Heroku

Heroku AI PaaS in Action

1. Solar Energy Dashboard:
   • Demonstrates a dashboard application that integrates with a PostgreSQL database deployed on Heroku
   • Uses Heroku's managed inference and agents to access the database and perform data analysis tasks, such as finding the lowest production day or charting hourly production data
   • The agent uses Heroku-provided tools and libraries to securely access the database and generate insights without exposing the database credentials
2. AWS re:Invent 2025 Agenda Builder:
   • Showcases Heroku Bots, the natural language-based application builder
   • Allows creating an agenda builder application for AWS re:Invent 2025 by providing a natural language prompt
   • The agent generates a step-by-step plan, which the user can review and approve
   • The agent then proceeds to build and deploy the application directly to Heroku

Key Takeaways

• Heroku aims to provide a Heroku-like developer experience for building and deploying AI applications and agents
• The Heroku AI PaaS offerings simplify the complex decisions and operational challenges involved in building AI-powered applications
• Heroku's managed inference and agents service allows easy provisioning and integration of curated AI models
• The vector database and Heroku Bots tools further enhance the AI development experience on the Heroku platform
• Heroku's goal is to increase developer productivity, reduce DevOps and infrastructure costs, and improve the return on investment for AI-powered applications