The editorial argues that Riddle's real significance isn't the Harry Potter reference but its demonstration that e-ink inverts every constraint that makes laptop/phone chat interfaces distracting: slow refresh, high contrast, single-purpose, no notifications. Latency stops being a bug when the surface is meant to be patient, making e-ink a uniquely well-suited medium for reflective LLM conversation.
Rivest built a working end-to-end pipeline that taps the reMarkable's stylus input stream, batches strokes, runs OCR, sends text to an LLM, and rasterizes the reply back into the tablet's native rM drawing format so it renders as ink. He's explicit that it's a hack requiring developer mode, but the engineering — using Linux/SSH access on the device — is real and reproducible.
The submitter framed the project as 'Fable turned reMarkable into Tom Riddle's diary,' and the post drew 258 points and 148 comments largely from users posting variations of 'I need this.' The organic enthusiasm suggests reMarkable owners have quietly wanted exactly this kind of ambient AI interaction on their tablets, but the market hasn't shipped it.
Maxime Rivest published a repo called Riddle that does something most reMarkable owners have quietly wondered about: it turns the tablet into Tom Riddle's diary. You write on the page with the stylus. A moment later, ink appears in a different hand and answers you. The reference is on the nose — the README says so — but the engineering is real.
The project hit the Hacker News front page with 258 points, mostly from people posting variations of "I need this." The demo video shows the loop working end-to-end: handwritten question, brief pause, handwritten reply drawn back onto the same page. No screen flash to a chat UI, no phone, no keyboard. Just paper that talks back.
Under the hood there is nothing exotic. The reMarkable runs Linux and exposes SSH. Riddle taps into the stylus input stream, batches the strokes, runs them through OCR, sends the text to an LLM, and then rasterizes the model's reply back into the tablet's rM-native drawing format so it renders as ink rather than as text. Rivest is clear that it's a hack — the code is small, the setup instructions assume you're comfortable enabling developer mode on a $400 device, and there is no polish layer.
The interesting thing about Riddle isn't the Harry Potter cosplay — it's that e-ink is quietly becoming a credible surface for LLM interaction, and almost nobody is building for it. Every chat product on the market optimizes for the wrong constraints for a lot of thinking work: instant tokens, colorful UI, a device that also holds Slack and TikTok. E-ink inverts all of that. Refresh is slow, contrast is high, the device does one thing, and there is no notification lane. Latency stops being a bug when the surface itself is meant to be patient.
That's the shift Riddle stumbles into. Chat with GPT or Claude on a laptop and you're in a fight for attention with the rest of the OS. Chat with a model through a reMarkable and the model is the only thing on the page. The tablet's slowness — normally its worst review-bait — becomes a feature. You write a paragraph. It writes a paragraph. Nobody is racing. The Riddle demo is the first time an LLM interface has looked like a journal instead of a terminal.
There is also a stack story here that's worth naming. The pipeline is: stylus input → stroke capture → OCR (this is the hard part on messy handwriting) → prompt assembly → LLM call → response tokenization → vector-ink rendering → push to framebuffer. Every link in that chain has open-source tooling now. Two years ago the OCR step alone would have gated the whole idea; today handwriting OCR is good enough that a hobbyist can wire it into a working demo over a weekend. The bottleneck has moved from perception to product design.
Community reaction on HN split predictably. One camp treated it as toy — cute, but reMarkable's own AI features (the "Type Folio" integrations, the recent handwriting-to-text improvements) will fold this into the product within a year. The other camp pointed out that reMarkable's official roadmap has consistently under-shipped on anything LLM-shaped, and third-party hacks like this have historically been the only way interesting things happen on the platform. Both are probably right.
If you own a reMarkable and you're comfortable with SSH, Riddle is a two-hour Saturday. The dependencies are small, the repo is readable, and swapping the LLM backend is a one-file change. Point it at Claude, GPT-4o, a local Llama, whatever — the interface contract is just "text in, text out." The rendering step is where the actual craft lives, and Rivest's implementation is a good reference for anyone thinking about e-ink output more generally.
More broadly, if you're building AI products, this is worth staring at for a minute. The dominant AI form factor right now is a chat box on a screen that also shows your email, and every product decision downstream of that inherits the assumption that the user is distracted. E-ink is one of the few consumer surfaces left where that isn't true. Kindle Scribe, reMarkable 2, Boox, the new Daylight Computer — the install base is small but the users are self-selected for wanting to focus. Nobody has shipped a serious LLM product for any of them. The first team that does will not be competing with ChatGPT; they'll be competing with a Moleskine.
One caveat worth being honest about: this is a hack, not a business. reMarkable's official firmware updates have a history of breaking hobbyist tooling, and the OCR quality on cursive or bad handwriting is still the tax you pay. If you deploy Riddle on your daily-driver tablet, expect to re-flash it at some point. The value here is what it points at, not what it ships.
Expect reMarkable, Amazon, and Boox to all ship a first-party "ask the page" feature within 12 months — the demand signal from projects like this is too loud to ignore, and the underlying tech is now cheap. The more interesting question is whether any of them ship it as a *journal* rather than as a chat window bolted onto a stylus. Riddle's real contribution is showing what the good version looks like: no cursor, no send button, no model badge in the corner. Just ink, waiting for a reply.
Ironically, Tom Riddle's diary encouraging kids to commit suicide [1] and then showing them how to do it would be more on brand than the version in the books.[1] https://en.wikipedia.org/wiki/Deaths_linked_to_chatbots#Suic...
Why so many negative comments on this? I find it cool. I showed this to my son and he absolutely loved it. The author is clearly a fan and this by no means is a production-grade product. Just someone tinkering in their free time and gave the DIY guide away for free. I just don’t understand the negat
I understand that this is kind of beside the point, but it seems like a bad thing to compare inventions to haunted artifacts that mind controlled their users into betraying their friends to a powerfully evil being. (Though since it is being powered by GenAI, which has also driven people to do bad th
This is sick, in the 90's Tony Hawk sense of the word.I love that people can just bang stuff into existence now.There were times in my life where I would wait for an engineering team to change the color of a button for a day to a week.We are not in the slow times anymore.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
In case anyone else wanted to get to a demo quickly:https://xcancel.com/MaximeRivest/status/2073544461473169432#...Original link for those who want it: https://x.com/MaximeRivest/status/2073544461473169432