Fable 5 being back really shows that using LLMs properly is genuinely a skill.

Since Fable 5 has landed back, I have been picking up where I left off and seeing fantastic results with the desktop app I have been building. What’s interesting though, I go on x and other sub reddits to see other people claiming “it’s been nerfed” “doesn’t feel the same” etc. it feels the same to me as it did before. Providing me with working solutions, checking in with our md files for plans and preferences. Most importantly, the communication is to the point and so much easier to read than 4.8.

Now full disclaimer, I work in tech(not a dev but work with them a lot) . So I can ask it to build according to a specific architecture pattern, data models, I can build API specs and give it to Claude to describe what I want changing or created. I know how to search from frameworks for it to use depending how on what I’m trying to build. I don’t always get 100% perfect results but it nails it 85% of the time. However it does feel like if fable doesn’t one shot a vague prompt, people are quick to call it nerfed.

I think knowing where to start on a project is pretty important to the outcome of any app you build with AI.