Developer Productivity

At the Pragmatic Summit, I heard firsthand that Uber engineers aren’t just using AI to write code anymore, they’re assigning it work. Let’s see how that plays out.

I was in the room at this year’s Pragmatic Summit when Laura Tacho dropped the numbers: nearly all developers use AI coding assistants, over a quarter of production code is AI-written - and yet productivity gains haven’t budged past 10%.

When teams don’t share information, clash on priorities, or follow unclear processes, even simple tasks can drag on for months.

We’re past just making better tools: machines will do the repetitive coding, and humans will focus on guiding, judging, and architecture.

Ex-Tesla AI director now programs mostly in plain English with AI, calling it the biggest workflow change in 20 years.

Software teams don’t implode because of bad code - they implode because someone really needed to win the argument about it.

The following is an attempt to clarify the circumstances of a production incident that happened more than 10 years ago; it is also a lesson to all of us to decouple our production servers from boilers in distant Balkan countries.

Developers still have agency, but now they also have smarter tools. The question is: are they using them wisely?

As AI coding assistants level up from autocomplete to autonomous, the real challenge isn’t what they can do - it’s how we use them wisely.

It’s no longer about fixing syntax errors - it’s about figuring out why a billion-parameter model did something totally unexpected.

Managing Python dependencies has always been a pain point - from slow installs to version mismatches across projects. uv aims to fix that.

Every team needs a hero, especially when support pings, bugs, and tech debt all strike at once.

Meet AgentKit, OpenAI’s drag-and-drop playground that handles all the boring LLM wiring for you.