#Links, quotes, and podcast episodes

I think the concept of erosion of trust is useful when thinking about how to use LLMs to contribute to team codebases.

Thoughts on choosing what to write about, how to say it, and the value of writing.

Transcript here.

A podcast introducing the DuckLake project with Hannes Mühleisen and Mark Raasveldt. Transcript here.

Thirty short reminders of how much is going right in the world, from nuclear weapons not destroying the atmosphere to the wonders of modern technology and medicine.

A service that takes any github repo and turns it into a docs site you can chat with.

Transcript here. I guess this could be used to store a db in s3 and use it to power a dashboard?

Ultra-cheap image segmentation feature and how to use it with the updated API. Tested myself and worked remarkably well.

Useful markdown rendering tool. Code here: GitHub repository for Simon's tools

Good lengthly article for LLM ingestion about using uv. See also the.dusktreader blog Self-contained Python scripts with uv

Good post about how to download bulk address or persons data from wikidata for e.g. synthetic data generation

A detailed walkthrough of using LLMs for both greenfield development and legacy code maintenance, with specific prompts and workflows for different scenarios.

Prompting can go very deep!

What are agents and how do we expect them to evolve.

A good guide to thinking about whether scaling is dead.

Not directly relevant to LLMs, but it's interesting to think at what point an LLM could produce an article like this. I feel like they're a long way off.

Hacker News discussion about companies moving from Apache Spark to DuckDB.

Interesting on interpretability

This article made me think of LLMs as like really software with no tests, no documentation, and lots of bugs. And yet very useful.

DuckDB's approach to handling concurrent transactions for analytics workloads.

The CEO of Anthropic outlines how AI could transform the world for the better.

There are lots of examples of strange capabilities like this you'd never see in a benchmark.