The pull request is dead. Code review isn’t. If your review process is “read every diff”, agents already broke it. If it’s “push to main and watch the logs”, that’s not a process. Reading an agent’s diff is checking a flight by its landing. You want the flight plan, the black box, and what the pilot learned on the last leg. More soon.

👨🏻‍🍳

Join the @EntireHQ live AI presentation over at Stage 5 at 11.40am with @jkcso, and learn how Entire is re-imagining the AI-native developer lifecycle. Don’t miss it 👀

Here at the best @WeAreDevs San Jose booth w/ @jkcso ! Stop by all day to see our demos, meet Marvin, && get merch! #wearedevs #wearedevelopers

Our friend @spedemo has made a real Marvin, 3D printed, remote controlled, with voice detection and speech, all open source. And it connects to ChatGPT and Entire. Come meet Marvin, Stefano and us @wearedevs San Jose, booth 753. 🌠

Who's at @WeAreDevs this week? @EntireHQ will be at Booth 753 (Hall 2) and present 3 sessions including a 2-hour workshop by my colleagues @lizziepika and @evisdrenova this Thursday. On Friday morning we follow with 2 demo-heavy sessions, one by Karthik Rameshkumar, and the other by myself showing how you should review code in the agentic era so you can keep up with it AND make it an enjoyable experience. Not everything you do has to feel like it's 2024. Come speak to us and join our sessions!

Agents are fast. Reviewing… not so much. So how do we change our workflow to adjust? Join @JosephKatsioloudes at #WeAreDevs North America for a workshop on solving the review bottleneck with context, not just diffs. 
Sign-up: wearedevelopers.com/world-co…

I’m using Astra daily now because of its speed, but I find Fable 5.1 more trustworthy for the hard problems. There was a case Astra was guaranteeing me for multiple turns it couldn’t resolve X because of Y, and Fable did it for me in half prompt. I’m thinking with Claude and executing with Astra.

have never written so many times the word yolo anywhere and my terminal looks like codex —yolo “this” codex —yolo “that” 😂

1) Are live demos better on terminal or the corresponding dektop app? 2) Speed-wise what’s the best model between those from @AnthropicAI and @OpenAI ? These are my current settings because Fable overthinks it even at low effort.

Conclusion: night coding beats morning coding.

Backstage from 🇳🇴 Oslo and @NDC_Conferences. Oh and I brought really cool @EntireHQ swag with me, time to get some Marvin stickers!

At the European Hackathon League in 🇨🇭 Zürich with @EntireHQ. SO amazing being among builders!

Astra is gooooooooooooood.