Maybe we're approaching AI engineering backwards.
Humans aren't deterministic either. We change our minds, contradict ourselves, and often don't know what we want until we see it.
What if AI could figure out what we need better than we can specify it?
Would we dare let go?
@AnthropicAI Yeah nice, facing a problem and solving it ππ».... But I guess for most people, companies, projects out there, git is still the better choice π
@elonmusk The real split isn't process vs. progress. It's process that enables progress vs. process that avoids decisions. Governments often reward the second.
Everyone's posting fancy agent graphics. Here's ours, built from git history and agent logs.
DATAMIMIC CE refactoring: 175 commits, 1,886 file changes, 23 β 9 package groups.
80 sub-agents in the last 5 days. 55 of them did nothing but verify.
Most agents checked. Few wrote.
@ohdearlordylord@grok Can you break this down for someone new to the topic?
What exactly am I looking at? How was this graphic created, and what data or assumptions drive it?
What are the main benefits, limitations, and trade-offs?
Keep it simple, practical, and objective. No hype.
@coryalthoff Everywhere obviously... Just vibe coded another trading platform with super fancy gif execute automatic trading ... Making 6.846.467,09 π² per day with magic data and beautiful visualization, only took me 2 h and the wish for getting rich , super fast
@Saanvi_dhillon I guess it's a good editor , use it , but most projects have python stack and I still like the Jetbrain products , maybe it's i am a bit nostalgic, like to feel close to the code π
@Taniyatweets_ First prompt: "Oh my God, this is magic." 60 billion tokens and 3 weeks of real data and real use cases later: the solution finally becomes useful.