Dagens Vibes — 22. juni 2026

Lenny Rachitsky@lennysan

Fiona Fung (@Nerdi_Yogi) leads the teams behind Claude Code and Cowork at @AnthropicAI (overseeing all of eng and PM, including @bcherny and @_catwu). Before Anthropic, she spent 11 years at @Microsoft building Visual Studio and TypeScript and then moved to @Meta, where she helped build Meta’s first VR and AR glasses, started Facebook Marketplace (now generating over $100B in GMV annually), and led @Instagram's infrastructure, growth, and safety teams. She’s been an engineer for over 25 years and has such a unique lens into where things are heading for product teams. In our in-depth conversation, we discuss: 🔸 Specific ways her team uses AI 🔸 The emerging context-switching and loneliness problem for engineers 🔸 How her teams do planning 🔸 What she’s learned about running teams shipping 8x more code 🔸 Which roles AI will transform next 🔸 What keeps her up at night Listen now 👇 https://t.co/2S79IakE9x

Peter Yang@petergyang

I am pretty skeptical of building loops for everything because I'm paranoid about the agents creating slop if I'm not there to monitor what they're doing. What are some good loops to try for coding or regular knowledge work?

Theo - t3.gg@theo

Want a good "first loop" to use with Codex? "When you are done designing the API, get a second opinion from Opus with 'claude -p'" This has significantly improved the quality of the code I get out of OpenAI models.

maria@maria_rcks

I went a bit too hard on looping a whole project start to end, it's beyond saving. Too many assumptions, no matter how detailed my instructions where for it to orchestrate other agents, and everything reviewed to hell and back (often by multiple agents) Now I have 34k LOC (almost a g unit), of code that’s misguided and not really the program i had pictured. And I think that’s okay. Maybe I just don’t have what it takes to be that detached from the creation process / birth of a project I’ll build this more closely next time, and try less to maximize some statistic. I’d rather build something I actually enjoy using, something more rooted in what I want it to be. Quality over quantity, I guess.

Citeret tweet

Theo - t3.gg@theo

@maria_rcks loop responsibly

♥ 89↻ 1💬 5🔖 1

Armin Ronacher ⇌@mitsuhiko

I decided to do some experiments with looping over the weekend. The only cases where they work so far for me are a) review b) research c) autoresearch. If someone uses them for actual implementation on a medium sized project, would love to have something to look at!

Chubby♨️@kimmonismus

Even the Vercel CEO is impressed/shocked at how good GLM-5.2 in coding is. open source, open weights.

Citeret tweet

Guillermo Rauch@rauchg

Genuinely impressed, almost shocked, at how good GLM-5.2 by @zai_org is at coding. This changes things.

♥ 6.6k↻ 357💬 239🔖 1.9k

Robin Ebers · AI for Small Business@robinebers

GLM 5.2 is *not* cheap $42.38 for a review workflow in Claude Code it burnt A LOT of tokens but most importantly, was it worth it? so I asked Opus 4.8 (Max) + GPT 5.5 (High) to evaluate the review most of it was useless a lot of was contradicting information, despite running through a verification workflow but there were some solid changes that lead to a PR (which I had Claude implement after filtering) my read on this: GLM 5.2 is not a bad model, but that it is nowhere near as good or as cheap as people make you believe the most interesting thing about this model is that it is pretty good, better than any other open source, and that it is truly open this will be the last time I used it

Chubby♨️@kimmonismus

Holy Sh*t: that changes the whole Fable 5 story completely: On June 11, the very same day Amazon reportedly uncovered the jailbreak, “Mythos” allegedly breached almost all classified systems belonging to the NSA and U.S. Cyber Command, not over the course of weeks, but within hours. "On June 11th Mark Warner, the vice-chair of the Senate Intelligence Committee, said that General Joshua Rudd, who leads the National Security Agency and the Pentagon’s Cyber Command, had told him that Mythos “broke into almost all of our classified systems, not in weeks, but in hours”." Via Economist

Peter Yang@petergyang

"I've spent $30K on a launch video, now you can make one with code." Here's my new episode with @liu8in and @JakeFromHeyGen, where they shared their 5-step playbook for using HyperFrames to make professional launch videos in Codex and Claude Code: 1. Gather video assets and create a frame.md 2. Create a storyboard.md with the scenes 3. Pull animations from the HyperFrames repo 4. Generate and review static frames 5. Build the video with HyperFrames Studio You can also start simple by using the built-in /website-to-video skill to one shot a launch video from any URL. HyperFrames is 100% free and I love how it democratizes video creation for everyone. 📌 Watch now: https://t.co/UPmXLJwbuO Thanks to our sponsors: @WisprFlow: 4x faster than typing with your voice https://t.co/22wo3B6qoT @RiversidedotFM: All-in-one AI studio for podcasts and video https://t.co/uWnS6aiMPE

Ben Vinegar@bentlegen

The future of code reviews will be multimodal https://t.co/9nHKsjSQDB