Postgres in the Age of AI (and Devin)

An AI software engineer was tasked with migrating a database from MongoDB to Postgres. Here's what happened.

#1about 2 minutes

Introducing Neon serverless Postgres for developers

Neon provides a serverless Postgres database that can be provisioned in milliseconds, offering a generous free tier for developers.

#2about 3 minutes

Using the AI engineer Devin to migrate from MongoDB

The AI software engineer Devin can automate complex and tedious tasks like migrating an application's database from MongoDB to Postgres.

#3about 3 minutes

Analyzing Devin's automated pull request for the migration

Devin automatically generates a pull request with multiple commits that convert queries, update dependencies, and normalize the database schema.

#4about 5 minutes

Prompting and observing Devin's step-by-step execution

A detailed look at the specific prompts required to guide Devin and its ability to plan, execute commands, and self-correct errors during the task.

#5about 2 minutes

Why AI agents require modern serverless infrastructure

Serverless infrastructure like Neon is ideal for AI agents because it handles mistakes cost-effectively and provides database branching for isolated preview environments.

#6about 3 minutes

The shift from developer experience to AI experience

The rise of AI agents is transforming developer experience (DX) into AI experience (AX), demanding faster provisioning and isolated testing environments from infrastructure providers.

#7about 2 minutes

How Neon's architecture enables the future of building

Neon's stateless architecture separates storage and compute, enabling the speed and flexibility required for a future where builders are humans augmented by AI.