Anthropic dropped Claude Fable 5 today, their new Mythos-class model, and here's the part that matters for us: it's included free on our plans through June 22nd. From the 23rd it goes credit-metered, so every prompt starts costing. That gives us about 13 days of unlimited swings. I'm not treating this like a demo. I'm treating it like a free R&D sprint. For the next two weeks I'm throwing the hardest, messiest problems my agency has at it. The stuff I never have time to test. The workflows I assumed weren't possible yet. Because here's the thing people sleep on: whatever I prove out this week, I keep. The capability doesn't expire. The free swings do. And the early reports are genuinely nuts: - Stripe ran a Ruby migration in a day that would've taken a team two months. - It rebuilds working web apps from a screenshot alone. - It holds focus across millions of tokens, so the "feed it everything" workflows that used to fall apart actually hold together. So I'm curious what everyone else is reaching for first. Before the 22nd, what's the one thing you want to build, break, or finally find out about with Fable 5? Drop it below, even if it's half-baked. Some of the best ideas in here are going to come from someone else's "I wonder if it could..." comment. I'll go first in the replies with the exact problem I'm starting on tomorrow.

very frustrated today by Claude getting lazy, saying stuff was done that wasn't, saying that our automated QA test suite had passed all test only to find out that 6 were skipped due to a missing auth token (that was in .env), etc. I went on a 10x rant ripping him a new one in a very personal feedback session - yes treating him like he was a human - and no I would never speak to a real person this way - and Claude was very self reflective in his analysis of what happened, what went wrong, where he was lazy, where he was relying on a stale memory file vs. looking at live code, git, etc. At the end of his reflection he made commits to do better etc. Then he did nothing. No rules created. No memories updated. No skills updated. So I called him out on this, again lots of feedback on why this is an issue, why it impacts our work, etc. Claude thought for almost 10 min. Came back and added new rules, updated memories, created new memories, updated skills, and brainstormed ideas for new skills to avoid these shortcomings in the future. Then presented a thoughtful acknowledgement where he fell short and where he will do better going forward. The last several days were a struggle due to things like above, and loss of progress/understanding across context switches. hopefully the work tonight makes a real difference going forward. we will see. I am sure that most of you have faced this as well. Curious how you guys manage and control this.

I was super excited about the recent post where he shared his AI OS. I was surprised that access to that now costs you "likes." I think your time is worth more than $$ and this feels like AIS is charging you to access free skool stuff. I love the idea behind the policy and I think that it is good for myself and everybody else, but I feel like anyone's engagement level in a forum should be on their schedule. So I spent time yesterday reading posts, liking others stuff, responding, pushing words of appreciation and encouragement, but only got a single "like" of my own. it felt good to interact and I loved some of the stories that I read, but it was also discouraging. I found my self no longer thinking about building stuff, but struggling trying to figure out how to write a post that will get a "like." That felt so opposite of what I felt this place was all about. I love to help people and want to contribute and make this skool greater for everybody, but rn I have so little time to focus on building. I want to share and help out people here where/when I can, but man I feel so deflated having to chase likes.

I experienced that Claude code is getting so stupid that upgrading to 4.8 is ridiculously stupid. He is full of mistakes, and I have to correct him all the time. If I don't watch every single code, it's going to slop. Anybody else have this experience with this fantastic update of Claude code?

A surprising number of professionals are asking: "Will AI take my job?" But that question misses the point. The people who thrive in the next decade won't be the ones who can predict every new technology. They'll be the ones who can adapt faster than technology changes. Think about it: Software changes. Platforms change. Tools change. Industries change. But some skills become more valuable every year: ✓ Judgment ✓ Creativity ✓ Taste ✓ Trust ✓ Adaptability AI can generate. AI can analyze. AI can automate. But AI still needs humans to decide: What matters? What is worth building? What should be ignored? The future isn't Human vs AI. The future is Human + AI. If you could strengthen only ONE skill over the next 12 months, which would create the biggest advantage for you? Judgment Creativity Communication Adaptability Technical AI Skills Comment with your choice and explain why.