Blake Rowley 👋🏽

@round What are you saying here?

design is so hot right now because it’s fucking cooked

@ValKatayev Na

Is OpenAI becoming Yahoo to Anthropic’s Google?

@letjudycook I still use it

old enough to remember Claude Design

@sikecannonbooks Or make a post on LinkedIn about the grass and what it told you about b2b SaaS

Every single Australian in tech needs to touch fucking grass. I went for a run this morning in the most beautiful country in the world get the fuck off linkedin.

@laraavci_ I love it but most product magic comes from user friction in repetitive patterns - that’s the hardest thing to overcome in interactions with expected defined paths

I think a lot of product magic will come from small new interaction patterns that make old UI assumptions feel weird in hindsight.

@tbpn Sometimes the sentiment is more important than the product. Agame bwcuase Figma really is a great product

$FIG just launched the Figma Design Agent, a dedicated agent that works alongside users in Figma. The agent lives directly on the canvas, so you can prompt from any design layer, iterate and edit alongside the agent, and hand it boring tasks that would have otherwise taken up a lot of your time. $FIG CEO @zoink came on the show today to discuss the launch. Here's our full conversation. 01:18 – Clocking AI-generated designs 03:31 – Figma’s new Design Agent 05:40 – AI code bloat and slop 09:09 – Using AI to augment creativity and branding 15:08 – Why Figma should be the front door to building apps in the AI era 18:53 – How design becomes more important as code commoditizes

@bchesky Really interesting to hear your more vulnerable take on the future of curation, discovery and aggregation UI too as a problem not solved. I think this is a great evolution nonetheless

Introducing the Airbnb 2026 Summer Release

I’m waiting to see how good the Gemini harness is before passing judgement

@rankintweets Not the first ragebait post I’ve seen saying the same thing… again.. tell people what they should instead of empty insult

Are people seriously still using pi and codex? lmao

@thsottiaux A bit of positivity never hurt anyone

What happened to Theo. Who did this to him

@theo I’ve found that per project instructions being a file you have to create , rather than a text box within the start of a chat to be a bit of friction but not something that can’t be overcome.

Honestly I'm still really impressed with the Codex app. It works reliably. It adds useful features consistently. It has taste. The mobile integration is awesome. The git integration is solid. If you haven't used it yet, I highly recommend it.

Original inspired article: x.com/trq212/status/…

Based on @trq212's excellent post called The Unreasonable Effectiveness of HTML (in 🧵).. I had some thoughts... I think the Word doc is dead. The next era of knowledge work won’t run on static files. It will run on live, governed HTML objects.

@karpathy @BjoernSchotte They’ve got a real superstar in their team, congratulations!

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

@AlfieJCarter AUTOPILOT

I just built a fully mapped Claude Code on Autopilot system covering every layer a GTM engineer needs to run autonomous builds from one connected task board. Feed it your project spec, your build requirements, and your current Claude Code setup → it maps every Linear issue, GitHub branch, and Slack notification to the right layer → so you know exactly what to connect, what each thing does, and how the layers keep Claude working without you prompting between tasks. All inside one free reference document. Perfect for GTM engineers and RevOps leaders who are still prompting Claude between every task, losing all build context when they switch devices or close a session, and manually tracking what was built and what still needs doing across multiple projects. If you're building GTM systems on Claude Code in 2026, you already know the math - the engineers shipping the most aren't prompting more, they're running a connected task board where Claude picks up the next issue, builds it, marks it done, and moves on without a single additional prompt. Most engineers use Claude Code without a second brain. No task board means scope drift, context loss, and mental overhead every session. This system solves it: → Linear connected in under 5 minutes - full read and write access so Claude sees teams, projects, and issues, updates statuses, adds completion comments, and moves tasks between columns automatically → CLAUDE.md rules file with exact behaviour instructions - read the assigned issue before touching code, only work on the current issue, update status on start and completion, create a GitHub PR before moving on → One-prompt spec system that generates your entire project as a Linear task board before a line of code is written - projects, issues, priorities, acceptance criteria, and sequencing all created automatically → Autonomous run setup - two prompts start the build, Claude completes 15-20 tasks continuously, your job shifts from builder to reviewer approving clean diffs → Multi-agent framework - Codex alongside Claude Code on the same Linear board, each claiming different issues, writing to separate GitHub branches with zero overwrites → Slack visibility layer - #linear-updates for every status change, #github-updates for every PR created or merged, nobody asks "what are you working on" ever again No re-explaining the project every session. No losing track of what was built across devices. No manually deciding what Claude should work on next when the board already knows. What you get: - Linear connection guide and CLAUDE.md rules file for consistent autonomous behaviour - One-prompt spec system that generates 40-100 issues before a line of code is written - Multi-agent setup running Codex and Claude Code simultaneously on the same board - GitHub branch-per-issue workflow with sub-agent diff review before every commit - Slack integration covering Linear status changes and GitHub PR events in real time Built from weeks of running autonomous Claude Code builds across GTM projects with zero manual prompting between tasks. Want it for free? > Like this post > Comment "AUTOPILOT" And I'll send it over (must be following so I can DM)

@peer_rich I’m stunned it hasn’t been eliminated completely already

It's not only you–it's all of us!

This Mini Shai-Hulud wave is pretty serious right!?

@lennysan Such an awesome article!! I think it’s going to change more than just output… what’s the need for docs anymore?

HTML is the new markdown

Claude design double tokens.. yes pls

@trq212 @lennysan Such a good article that I actually created a video on the topic! @trq212 thanks for sharing the insight!! youtu.be/r_qPDBTgelc

continuing my HTML era, I had so much fun talking with Claire at Code w/ Claude about staying in the loop with long running agents