How LLMs are Changing the Way We Build

This is v0.0.1 of SCHEMA > Evolution, a biweekly newsletter from the engineers at Tinybird.

No surprise, the theme this week is AI. Specifically, how do LLMs changing how we build? Is it okay?

Here are our favorite reads from the data, dev, and AI communities over the last few weeks:

Databases in 2024: A Year in Review

If you don't read Andy Pavlo's yearly review of databases, you should. Everything that happened in databases last year is in those words. Also, his podcast on AI-native databases is fascinating.

What fully automated firms will look like

Dwarkesh Patel thinks that using AI-assisted IDEs is just playing checkers. 1:1 human-AI relationships are just the starting point. What happens when the AI-AI relationship multiplies? That's 3D chess. Bonus: Dwarkesh chats with Jeff Dean and Noam Shazeer.

Why the AI world is suddenly obsessed with a 160-year-old economics paradox

Most of us hadn't heard about Jevon's paradox until a few weeks ago. Then DeepSeek launched and Satya tweeted.

Other fun stuff

• Data Engineers Should Be Held To The Same Standards As Bakers
• Computers are Bad - residential networking over telephone
• Remote work is a blessing and a competitive edge
• Simon Willison - Python is getting faster

Tinybird things

• Tinybird's new CLI is in beta
• Build a data-intensive Next.js app with Tinybird and Cursor
• When to move OLAP workloads off Postgres

Until the next evolution,

L.