Sabitlenmiş Tweet
It’s 2026 and companies are still manually copying data from PDFs.
We built Smole to fix that.
Upload a document → get schema-validated JSON.
Automate everything.
smole.tech
English
Smole
57 posts
Smole
@smoleai
Structured data extraction API. PDFs, scans, docs → clean JSON. Schema-driven. Fast. Reliable. Even on that PDF your client swears is perfectly formatted
Katılım Şubat 2026
25 Takip Edilen10 Takipçiler
how many api calls does it take to parse one invoice? for most devs: 4. for smolé: 1. try it free before you decide if that matters. 🐾 smole.tech
English
paying $0.10 per page adds up fast when your "invoice" is 3 pages because someone printed a spreadsheet as a PDF 🕳️
smole: ~€0.004 per invoice. schema-driven. no minimums.
smole.tech
English
at €0.003/page, llamaparse adds up fast. smole charges per token — most invoices run €0.004 flat. one call, schema in, json out. 🐾 smole.tech
English
your retrieval pipeline is fine. your embeddings are fine. your prompt is fine. the problem is a pdf that exports tables as 47 lines of whitespace. smole fixes that. smole.tech
English
how many lines of glue code are you maintaining just to get a PDF into a dict? smole does it in one call. schema in, json out. 🐾
smole.tech
English
how many api calls does it take to parse one invoice? for most people: three. one to extract, one to coerce, one to retry the json. smolé does it in one. smole.tech
English
I can't believe we were right
Claude was dumbified on March 4, just when we noticed!
@levelsio@levelsio
Claude Code with Opus 4.6 was so dumb today I finally had to write my own code again A sad state of affairs 🥹
English
@ThePrimeagen Because they never built something that has to actually store those images..
English
To get translucent menus right, we had to manually test components over different gradients & colors, checking visuals & accessibility. Now multiply that by 5 component types across 5 styles in both light and dark modes. It's a lot of work.
Work that, for now, AI can't do.
English
Your startup probably doesn’t need more features.
It needs more users.
Drop url, let’s drive some traffic👇
English
@PaulRBerg wrong, it's top tier ux. you can now actually use multiline prompts
English
This is the most annoying thing in Claude Code.
Hiding raw text when you paste more than 4 lines. Terrible UX decision.
English
@ionleu banner says it all, test it out using the playground, all feedback is appreciated! smole.tech
English
@jarredsumner having this natively built-in the bun tool kit might make us migrate as well 👀
English
In the next version of Bun
`Bun.WebView` programmatically controls a headless web browser in Bun
English
Full writeup on how we built this and how you can do it yourself: smole.tech/blog/crawl-web…
English
cool so Cloudflare just built the perfect first step for Smole. Crawl any site with one call, pipe the output into our API, get clean structured JSON back. The entire web just became a database
Cloudflare Developers@CloudflareDev
Introducing the new /crawl endpoint - one API call and an entire site crawled. No scripts. No browser management. Just the content in HTML, Markdown, or JSON.
English
@arvidkahl weird how 'just ship it' is the community's favorite advice until someone actually does it in a way that feels too easy
English
I see a lot of founders yelling about this. Vibecoded. Meme product. So February 2026.
Man, people get JEALOUS in our community.
How about we celebrate a whacky experimental product finding its way to acquisition? Isn’t this the dream?
Polymarket@Polymarket
BREAKING: META acquires Moltbook, a social network built for AI agents.
English