@kerckhove_ts Do you mean an LLM agent?
Would simply commenting TODOs where you want to have changes and tell it to look at current git diff, or last commit if you commit them, and tell it to look at them be enough?
@owickstrom It's quite nice to do a POC of something you don't know well, just so you can decide if you want to learn more about it or not. And if you're not in a hurry, you can build bigger POCs casually on a phone while walking or something like that across multiple sessions.
It sucks to be laid off and spied on and... But that's basically what they were working on, right? Spying and manipulating their users, hoarding copyrighted materials,...
Today I published an interview with an anonymous Meta employee who has worked at the company for over a decade and wanted, for the first time ever, to let the world know how horrible it feels to be inside. https://t.co/eXXPbiQhmV
I love how the industry frames code review as being the bottleneck now that PRs are full of LLM-generated walls of useless comments and spaghetti code.
They want to convince us they will do more with less. Why didn't they just do what they say, then reduce the headcount?
The quoted paragraph could start with "... Multiple human teams open merge requests [...] at a rate no human team ever did."
Agents are excuses.
"... Agents open merge requests in parallel, trigger pipelines around the clock, and push commits at a rate no human team ever did. Git itself wasn't designed for that load, and bolting AI onto platforms not built for agents is the biggest mistake of this era. We're doing a generational rebuild of the underlying infrastructure to handle agent-rate work as the default. Git itself is being reengineered for machine scale. The monolith is giving way to modern, API-first, composable services. And agent-specific APIs are being built so agents can act as first-class users of the platform, not as bolted-on consumers of human-shaped interfaces. The value of this 100x scale infrastructure, and the reliability and performance it provides is much higher than the generation of infrastructure in the market today. ..."
https://t.co/SRIUKyAVdG
When I tried to convince a bunch of academics at a workshop recently about designing for agents vs designing for humans, they looked at me as if I just smoked a crack pipe.
Colleagues at work use FPGAs, ROS, and other stuff I never learned so it go me intrigued. I've found @ShawnHymel videos really cool to get a better idea of those things.
The web is ephemeral and content we consume barely has any lifetime or accountability.
I'm building Vertex (https://t.co/aXppYkqcy9), and it is meant to change that.
Imagine https://t.co/hCaCxSefxz but on demand.
@krismicinski I'm also wondering if the "baseline data" are collected in the same context (same roads, same duration, amount of traffic, weather, car price, ...).
@krismicinski You seems to say the data can be unambiguously interpreted. What if it's not about autonomous driving, but just about having lidars, or foreign workers watching the camera feed, or people being aware there's a car with no driver in it going their way,...
What I want from a CI system: being able to run the same exact same thing locally, being able to run/watch/grep a remote run from my terminal, easily search build artifacts (including logs) possibly based on git references.