No More Copy-Pasting Into ChatGPT: This Open-Source Whiteboard Reads Your Handwriting
It's called PenEcho. You sketch equations, draw diagrams, scribble notes wherever they make sense. When you pause, it sends that region to the model. The response appears beside your work. Not in a chat window.
On the whiteboard itself.
The canvas is 20,000 x 20,000 pixels but only loads tiles where ink exists. Each request costs a few cents. Runs locally. Supports Claude, OpenAI, and Codex.
The builder said something that stuck with me: six months ago, no model could have done this. Not the handwriting recognition.
The fact that it understands unfinished equations, rough diagrams, and the spatial relationship between them. It infers what you meant from incomplete marks.
Link: https://lnkd.in/e9BUnSE2
This is how research should work. Think on a whiteboard. AI thinks with you. No copy-pasting into a chat box. No breaking your train of thought.
PenEcho is real — confirmed. Open-source (AGPL-3.0, ~768 JS files, actively maintained), on GitHub as penecho/penecho, with a companion site at penecho.ai.
Core mechanic (handwrite → region sent to model → response appears beside your work on canvas, not chat) — confirmed, matches the project's own description almost exactly.
"20,000 x 20,000 pixels, only loads tiles where ink exists" — confirmed. The project explicitly describes sparse tile rendering for the 20,000×20,000 logical canvas so it doesn't allocate one giant bitmap.
"Runs locally" — true but worth a caveat: the app runs locally (desktop app or Node.js package), but unless you're using a local CLI connection, inference still goes out to an API (OpenAI/Anthropic/etc). Not misleading, just slightly underspecified.
"Supports Claude, OpenAI, and Codex" — accurate but incomplete. It also supports Kimi K3, and PenEcho is actually an official member of Moonshot AI's "Kimi Open Source Friends" program, which backs it with API credits — Kimi K3 is one of its recommended models, not an afterthought. Leaving that out understates the model support.
"Each request costs a few cents" — plausible given per-region API calls, but I couldn't independently verify this exact figure from the project's own docs.
The builder quote ("six months ago, no model could have done this...") — I can't verify this was actually said by the builder. Attributing an unverified quote to a named/implied source is the riskiest line in the post to publish as-is.
Bigger issue than any single fact: this post is a near-verbatim match to a viral X post by Vaibhav Sisinty from mid-July, which itself credits the discovery to "someone on Reddit." Your version drops "on Reddit" and reframes it as "someone just open-sourced," which makes it read as your own fresh discovery/commentary rather than a repost. That's worth fixing before this goes out — either quote/credit the original, or rewrite it in your own words rather than lightly editing someone else's post.