Now consider the risk of legacy SaaS companies laying off a high percentage of their engineering staff, who have the full domain knowledge used to build their products.
The better play might just be carrying a lot more Bigheads on the Hooli payroll.
Figma shipped a silent patch specifically to kill figma-use — my open-source tool that did what they wouldn't: an MCP server that creates and modifies designs, JSX export, design linting. Then they scrambled to catch up with their own MCP server.
So I spent the weekend recreating @Figma from scratch.
OpenPencil: reads and writes .fig files, AI chat with full design tools, P2P collaboration with zero servers, ~7 MB app. No account, no subscription.
Three days, one developer, MIT license.
https://t.co/bPtP6JPbq0
@mikedemarais Where it shines for me is it can act based on any external event: a cron job, an every thirty minute heartbeat, an incoming email, a text from a friend. They’re now all inputs to an AI personal assistant. And I can set up a new automation w/a simple message from anywhere.
@im_roy_lee Virality is reproducible product-led growth. What you’re describing is a media bump.
The former has compounding benefits to a product. The latter is just short-term spurts of attention that need to be remanufactured every time.
If I hit this dictation button after sending a text and it beeps and stops my music one more time,
I’m gonna find everyone at apple and put them in a rear naked choke hold
Even if I turn off dictation I somehow hit the voice note thing
The send button should not have multiple functions in the same spot
The best operators are too busy building to broadcast. LLMs shrink the opportunity costs of teaching while doing.
If your community rejects AI-assisted writing, it’s not protecting quality — it’s choosing irrelevance.
@steipete Such a weird line to draw. What about spell check? Google translate? Grammarly, Textio? Evernote? Voice to text?
Writers of all levels publish on the internet, and slop is not exclusive to LLMs.
The value of the writing should be judged on its own merits.
One of the core features of most languages is human ergonomics, but that’s also what leads to lots of ambiguity for ai (implicit returns, function overloads, type coercion).
Thought experiment: if you were designing an ideal programming language to work w/the shortcomings of today’s AI Agents, how would it differ from programming languages written for humans?
@steipete@koeckc Yep, just runs in your own env, no data going to Anthropic’s cloud. Big company stuff… but its metered token usage only so costly. Advantage for those that can ignore it.