Eric Haft

[Eric Haft @erichaft]

@petergyang @trq212 Nice! I’d imagine you’re just 1 or 2 prompts away from adding a commenting feature to those files that saves locally and AI can read, at least to POC it?

[Eric Haft @erichaft]

@thomasschulzz I guess my point is when you’re flaring on the feel you want, that’s fast and cheap with Chat GPT which excels at exploring styles with image generation 2.0. From there you hand the winning style over to your flow

[Eric Haft @erichaft]

@egsmrnv @thomasschulzz Right but when you’re iterating quickly to get the look you want, isn’t Chat GPT image generation going to better at flaring fast and cheap on style choices than building a font every iteration? Then you hand the winning image over as reference to build the font.

[Eric Haft @erichaft]

@thomasschulzz Maybe I’m confused. I generation a flat image of a font in ChatGPT using its image generation. Then Claude (or Codex, whatever) takes that flat image, and turns it into a font. Are you suggesting a different flow?

[Eric Haft @erichaft]

@zendadddy I’m with you. I used to be able to hack my way through frontend code and it was so exciting to bring your own designs to life. The coding world became too complex to keep up with. Now it’s unlocked again like it never was before. So much fun!

[Eric Haft @erichaft]

@PrajwalTomar_ What do you mean? It only “learned” what it wrote to memory and skills files when it comes to your specific project. Why can’t another model just pick up where it left off? I’m not really worried my projects are so unique that Fable is meaningfully different based on training.

[Eric Haft @erichaft]

@joshpuckett Touché, applying some real product thinking to it. Thanks. It IS super confusing -even the onboarding. The work tab has this suggested prompt “Review recent unread emails and draft personalize follow-ups” but I can give the same prompt to the chat tab, and get the same outcome 🤷

[Eric Haft @erichaft]

@designcoursecom Claude doesn’t have image generation. Why not use GPT 5.6?

[Eric Haft @erichaft]

Even though I know it’s there, I always have to think real hard to find my X bookmarks under “history”.

[Eric Haft @erichaft]

@dayonefoundry Does this just equate to the amount of energy cost we’re really burning? Hard to beleive

[Eric Haft @erichaft]

@krispuckett Why is my whole feed trolling today?

[Eric Haft @erichaft]

When flaring and ideating, does anyone ask AI to generate extremely low fidelity concepts? When asking it to “generate five ideas“ is there any value anymore in dialing back the fidelity at that early stage when it doesn’t save any time?

[Eric Haft @erichaft]

@cryptopia2222 @AnthropicAI Yeah @AnthropicAI @trq212 just do what me and this totally credible guy named cryptolambo say, thanks!

[Eric Haft @erichaft]

Is today the day where @AnthropicAI says it’s resetting limits and extending the fable 5 deadline? Please?

[Eric Haft @erichaft]

@rjs I’m reading your post again and I do agree. Someone does need to own it, and product needs to be informed enough to get opinionated with XD, eng, and others. But it cuts both ways and those roles need to understand the business value too and know when to push and be collaborative

[Eric Haft @erichaft]

@simonw @MengTo just shared a pretty impressive landing page, where the suggestion is the exact opposite and shows the very detailed prompt. Maybe Fable is the first model where you can start to get away with it? https://t.co/zAiFOjmdzu

[Meng To @MengTo]

We might be entering an era where writing an incredibly detailed prompt matters more than building the landing page itself.

With Fable 5 becoming this capable and this expensive, the prompt is quickly becoming the most valuable asset.

This interaction was one-shotted. The level of polish and interactivity is insane.

[Eric Haft @erichaft]

@ridd_design Purple gradient briefs 😀