The Short Answer
No single device closes the loop cleanly. The setup that actually gets handwritten notes into a searchable Obsidian vault with the least manual cleanup is a Boox Note Air-series tablet (Note Air4 C or Tab Ultra) running Android natively, using Boox’s built-in MyScript-based handwriting conversion to spit out plain text or Word files directly into a folder that syncs to your vault — because it skips the “export from a closed device, then convert, then import” chain that reMarkable and Supernote both force you into. If your priority is capture quality and reliable ink search rather than a unified vault, an iPad with Apple Pencil and GoodNotes 6 beats every e-ink device on the market and you should stop pretending Obsidian needs to own the handwriting layer at all. If you want the best writing feel and don’t mind typing your own summaries by hand later, reMarkable 2 or the Paper Pro is the nicest object to write on and the worst one to pipe into a markdown vault without a subscription and some scripting.
- OCR is the real bottleneck, not sync or export format. Clean block print converts at rates reviewers describe as “usable with light editing”; cursive and diagrams routinely fail outright.
- Boox devices run Android, so they can install Obsidian’s own app — but you still write in Boox’s native notes app, not inside Obsidian itself, because Obsidian has no ink layer.
- reMarkable’s handwriting conversion sits behind the $2.99/month Connect subscription; without it you’re exporting PDFs, not text.
- Diagrams, mind maps, and arrows never become text. The only honest path is embedding them as images in a note and tagging them manually.
- iPad + GoodNotes/Notability gives you ink search without OCR conversion, but that search lives inside the app, not inside Obsidian’s vault-wide search — you get a second, disconnected index.
- Every e-ink route to Obsidian is a pipeline you maintain, not a feature you turn on. Budget for that maintenance before you buy hardware.
The Three Routes From Pen to Vault
Everyone trying to combine e-ink and Obsidian ends up on one of three paths. They are not equally mature, and picking the wrong one for your handwriting style and patience level is why most people abandon the setup within a month.
Route 1: Android e-ink running Obsidian directly
Boox’s tablets (Note Air4 C, Tab Ultra, Note Max) run a customized Android 11 or 13 with full Google Play access. You can install the actual Obsidian mobile app on the device. This matters because it means your vault can live on the same physical hardware you write on — no cable, no cloud round-trip, no separate computer required to move files.
The catch: Obsidian is a plain-text markdown editor with no handwriting input of its own. Writing still happens in Boox’s native “Notes” app, which uses an on-device or cloud MyScript engine to convert strokes to text. What Route 1 actually buys you is proximity — you convert a page, export it as .txt or .docx into a folder, and Obsidian (already open on the same tablet) picks it up instantly if that folder is inside your vault. There is no live “type as you write” integration; conversion is a discrete, on-demand step you trigger per page or per notebook.
This route is for people who want one device, are comfortable doing light copyediting on converted text, and don’t need their diagrams searchable — only findable by note title or tag.
Route 2: Export and transcribe
This is the reMarkable and, to a lesser extent, Supernote model. You write on a closed device, then get the ink off it — either as a PDF export, a converted text file (behind reMarkable’s Connect subscription or Supernote’s bundled cloud OCR), or via third-party tooling like the open-source rmapi client that talks to reMarkable’s cloud API and pulls files down for local processing.
Once you have a PDF, you have two sub-options: trust the device’s own handwriting conversion (fast, mediocre on cursive) or run the PDF through a dedicated recognition tool such as Mathpix Snip or MyScript’s standalone apps, which reviewers generally rate higher on messy handwriting and mixed math/text pages than the conversion built into either reMarkable or Supernote’s first-party apps. Either way, the output is a text or Word file you manually drop into your vault, or a PDF you embed as an attachment and link from a markdown note.
This route is for people who like the specific writing feel of reMarkable’s Canvas display or Supernote’s Carta 1200 panel more than they dislike a manual transcription step, and who are disciplined enough to actually do that step regularly rather than let a backlog of unconverted PDFs pile up — which is the single most common reason these systems die.
Route 3: Tablet and app with handwriting search
This is the iPad route. Apple Pencil plus GoodNotes 6 or Notability doesn’t convert your handwriting to text at all by default — it indexes your strokes so the app’s internal search can find a word you wrote in cursive without ever generating a clean transcript. GoodNotes calls this “handwriting search”; it works directly against the ink, including a reasonable amount of cursive, and reviewers consistently report it as more forgiving than dedicated e-ink OCR on the same handwriting samples.
The honest trade-off: that search index lives inside GoodNotes or Notability, not inside Obsidian. You can export a PDF of a notebook into your vault and link it from a markdown note, but Obsidian’s own search will only ever see the note’s title, tags, and any typed summary you add — never the ink content. You end up maintaining two knowledge systems that don’t talk to each other, which defeats a chunk of the point of a PKM tool, unless you treat Obsidian purely as the index-and-link layer and GoodNotes as the archive.
Where OCR Actually Breaks
Almost every consumer handwriting-to-text engine on these devices — Boox’s, Supernote’s, and reMarkable’s — licenses or approximates the same underlying technology family, MyScript’s iink engine, with different tuning and UI wrapped around it. That means the failure modes are largely shared across brands, not brand-specific:
Cursive with connected letterforms. Print-style handwriting with clear letter separation converts at rates most reviewers describe as good enough to use with minor corrections. True cursive, especially fast, low-baseline-discipline handwriting, drops noticeably — expect frequent word-level errors that require re-reading the source image to fix, not just spell-check.
Diagrams, arrows, mind maps, and non-linear layouts. None of these convert to anything at all. The engines are built to recognize a left-to-right, top-to-bottom writing flow. A boxed diagram with arrows pointing between nodes gets either ignored or garbled into nonsense text fragments. The only reliable move is to leave these as embedded images and rely on manual tags for retrieval.
Math notation. Superscripts, fractions, and multi-line equations are a separate problem domain that general handwriting OCR doesn’t attempt to solve. Mathpix exists specifically because MyScript-class engines don’t handle this well; if your notes are math-heavy, factor Mathpix Snip into the budget rather than expecting your e-ink device’s built-in conversion to cope.
Mixed-language pages and technical vocabulary. Engines trained on general-purpose dictionaries mis-recognize domain jargon, code snippets, and proper nouns at a noticeably higher rate. If your notes are dense with specific terminology, expect more correction time than a demo would suggest.
Baseline drift and page pressure. Long free-writing sessions on any e-ink surface tend to drift diagonally across the page as your hand loses its reference point, and OCR accuracy degrades as drift increases — a known, structural issue with unlined or lightly-lined templates rather than a bug in any one device’s software.
The practical implication for a PKM system: treat OCR output as a rough first draft you skim and correct, not a transcription service you trust unattended. Anyone selling “seamless handwriting to searchable notes” as a finished feature is glossing over this.
Device Comparison
| Device | Screen / PPI | Pen pressure levels | OCR quality (reviewer consensus) | Export path to Obsidian | Sync friction | Price |
|---|---|---|---|---|---|---|
| reMarkable 2 | 10.3″ E Ink Carta, 226 PPI, 21ms latency (manufacturer figure), no backlight | 4096 (with Marker Plus; base Marker has no pressure sensing tilt) | Good on clean print, weak on cursive; conversion gated behind Connect subscription | PDF/handwriting export via Connect cloud, or third-party rmapi pull to desktop, then manual drop into vault folder |
High — no native folder sync, no official Obsidian integration | $399 device + $2.99/mo or $69/yr Connect; Marker Plus adds $149 |
| reMarkable Paper Pro | 11.8″ color E Ink, 229 PPI, front light | 4096, tilt-sensitive Marker Plus included tiers | Same conversion engine as RM2; color layer doesn’t improve text OCR | Same Connect-based export pipeline as RM2 | High, same as RM2 | $579 and up, Connect subscription still required for conversion |
| Boox Note Air4 C | 10.3″ color E Ink Kaleido, ~300 PPI mono / ~150 PPI color | 4096 with Pen Plus stylus | MyScript-based, on par with Supernote on print; cursive still weak | Native Android — install Obsidian app directly, export converted text into vault folder on-device | Low if vault stored locally or via Syncthing/Dropbox already configured on Android | ~$500, one-time, no subscription required for OCR |
| Supernote A5 X2 | 10.3″ E Ink Carta 1200, 300 PPI, no front light | 4096 | Neural OCR (cloud-based) rated by reviewers as cleaner Word-formatting output than reMarkable’s | Supernote Partner desktop/mobile app converts and exports .docx/.txt, then manual import to vault | Medium — no direct folder sync, requires the Partner app step | ~$429, one-time; OCR conversion bundled, no recurring fee |
| iPad (10th gen) + Apple Pencil + GoodNotes 6 | 10.9″ Liquid Retina LCD, 264 PPI, backlit | Not published as discrete levels; supports pressure and tilt | No conversion — ink search only, reviewers rate it more forgiving of cursive than device-side OCR | Export PDF, embed/link in Obsidian note; no text-level integration | Low for capture, but creates a second, disconnected search index | iPad from $349, Pencil $79–$129, GoodNotes 6 ~$9.99/yr |
| iPad + Notability | Same iPad hardware options | Same Pencil specs | Similar ink-search behavior to GoodNotes; audio-note sync is Notability’s differentiator, irrelevant to OCR quality | Same PDF-export/link pattern as GoodNotes | Low for capture, same fragmented-index trade-off | ~$2.99/mo or $29.99/yr; legacy one-time purchasers grandfathered |
Device-by-Device Breakdown
reMarkable 2 and Paper Pro
reMarkable still sets the bar for the physical feel of writing on E Ink — the low-latency, paper-textured Canvas display is what most reviewers point to when they say a device “disappears” under the pen. That is a real, meaningful advantage if you write a lot by hand and hate the slight lag other panels introduce.
It is also the most locked-down device on this list relative to Obsidian. There’s no file system you can browse the way you can on Android; getting content out requires either paying for Connect (which unlocks handwriting-to-text conversion and unlimited cloud storage) or using third-party command-line tools against reMarkable’s cloud API, which is unofficial and can break with firmware updates. If your goal is a tight Obsidian pipeline rather than a beautiful writing surface, reMarkable is buying you friction you’ll pay for monthly.
Not for: anyone who wants their vault updated same-day without a subscription or scripting step. Good for: people who mainly want a paper replacement for long-form writing and are willing to treat Obsidian import as a weekly batch chore.
Boox Note Air4 C and Tab Ultra
Because these run stock-ish Android, they sidestep the closed-ecosystem problem entirely. You can install Obsidian, a file manager, Syncthing, or Dropbox, and treat the device as a small Android tablet that happens to have an E Ink panel. The color Kaleido panel is dimmer and lower-contrast than the mono panels on reMarkable or Supernote, and text sharpness at 150 PPI in color mode is visibly softer than the ~300 PPI mono reading experience — a real trade-off for the convenience of Play Store access.
Handwriting conversion uses a MyScript-derived engine similar in accuracy to Supernote’s, which puts it ahead of casual expectations for print but no better than anyone else on cursive. The advantage isn’t OCR quality — it’s that the whole pipeline, capture to converted text to vault folder, can happen on one device without a cable or a second app ecosystem.
Not for: readers who want the crispest possible E Ink text rendering — dedicated mono readers still win there. Good for: anyone who has already built an Obsidian vault around Android-friendly sync (Syncthing, self-hosted WebDAV, or a synced cloud folder) and wants handwriting capture inside that same environment.
Supernote A5 X2 (and Manta)
Supernote’s Carta 1200 panel runs at 300 PPI with no color option on the standard A5 X2 (the Manta adds color and a front light at a higher price). Reviewers commonly rate its handwriting conversion output as cleaner when exported to Word — better preservation of line breaks and paragraph structure than reMarkable’s equivalent — which matters if your endpoint is a lightly-edited .docx rather than raw plain text.
The Supernote Partner app (desktop and mobile) is the conversion and export hub; there’s no native folder sync into an Obsidian vault, so you’re still manually moving files after conversion, but there’s no recurring subscription required to do the OCR step itself, unlike reMarkable’s Connect gate.
Not for: people who want same-device, zero-transfer capture — you’re always exporting to a separate app first. Good for: people who write dense outline-style notes and want the OCR output to preserve structure well enough to paste directly into a note with minimal reformatting.
iPad plus Apple Pencil (GoodNotes 6 / Notability)
This is the pairing to pick if what you actually want is reliable retrieval of things you wrote by hand, not a unified plain-text vault. GoodNotes’ and Notability’s ink-search features work against the stroke data itself, without converting anything to text first, and both are commonly reported as handling cursive and mixed print/cursive pages better than the dedicated OCR built into e-ink devices — because they’re solving a narrower problem (find this word in this notebook) rather than the harder one (produce a clean transcript of this page).
The cost is architectural: Obsidian never sees the ink. You can link a PDF export into a note and tag it, but full-text search across your handwriting only happens inside the notes app, not inside Obsidian’s own search bar. For people who treat Obsidian as a hub of links and typed synthesis rather than a single source of truth for every word they’ve ever written, that split is fine. For people who specifically wanted “search my handwriting from inside my vault,” it’s a dealbreaker, and no amount of tagging discipline fully closes that gap.
The Obsidian Side: Plugins, Sync, and Where People Get Stuck
Obsidian itself is free for personal use; Obsidian Sync is a paid add-on (roughly $4/month billed annually) for official cross-device sync of your vault, and Obsidian Publish is a separate $8/month product for web publishing — neither is required to make any of the above pipelines work, since plain folder sync via iCloud Drive, Dropbox, OneDrive, or self-hosted Syncthing does the job for markdown and PDF files.
There is no official plugin that accepts raw ink strokes and converts them inside Obsidian. What exists are workarounds: the Excalidraw plugin lets you draw directly inside a note with a stylus on an iPad or Windows tablet, storing the drawing as an .excalidraw file with an optional text layer, but its handwriting recognition is meaningfully weaker than GoodNotes or MyScript and it’s built for diagramming, not long-form handwritten note-taking. Plugins like Obsidian Git or Remotely Save handle syncing the resulting markdown files across devices but do nothing for the handwriting-to-text step itself.
The practical architecture that works for most people combines a synced “Attachments” folder for PDFs and images (the raw ink, kept as visual reference) with a synced “Inbox” folder where converted text lands, which you then manually review, clean up, and file into permanent notes with proper links and tags. That manual filing step is unavoidable in every route above — nothing on the market currently auto-files converted handwriting into the right place in a linked-note system, and treating any tool’s marketing as if it does will waste a weekend.
Who Should Just Use an iPad
If your handwriting notes are primarily for your own review — lecture notes, meeting notes, reading annotations — rather than for feeding into a searchable, cross-linked knowledge base, an iPad with GoodNotes or Notability is simply the better tool, and trying to force that workflow into Obsidian’s vault will cost you more setup time than it returns. The iPad’s LCD panel is backlit, runs at higher effective refresh than any E Ink device (E Ink’s black-and-white refresh in the 200–450ms range for full-page redraws is a hard physical limitation of the panel type, not a software choice), and both GoodNotes and Notability now support typed-text mixed with ink, PDF annotation, and multimedia in ways no E Ink device attempts.
The reason to choose E Ink anyway is specific: battery life measured in weeks rather than a day (reMarkable and Supernote both publish multi-week standby figures under light use), a non-backlit or minimally-backlit display that a meaningful number of users find less fatiguing for long writing sessions, and the absence of notification pull that comes with a full iPadOS environment. If neither of those matters to you more than search quality does, buy the iPad and stop optimizing for a PKM feature you won’t end up using.
Recommended Setups by User Type
Student transcribing lecture notes into a course-linked vault: Boox Note Air4 C, native Obsidian app installed on-device, handwriting converted in short daily batches rather than left to pile up, diagrams re-drawn digitally in Excalidraw when they matter enough to be searchable.
Journaler who wants the writing experience above all else: reMarkable 2 with the base Marker (skip Marker Plus unless you specifically want tilt shading), Connect subscription accepted as a cost of doing business, weekly export-and-file routine into Obsidian rather than expecting same-day integration.
Researcher annotating PDFs and papers heavily: iPad plus Apple Pencil plus GoodNotes 6, PDFs linked into Obsidian notes as attachments, Obsidian used purely as the index and synthesis layer, GoodNotes accepted as the permanent home of the annotated originals.
Outliner who writes structured notes rather than freehand prose: Supernote A5 X2, relying on its cleaner Word export to paste structured text into Obsidian notes with minimal reformatting, accepting the extra Partner-app export step as the price of no subscription.
Frequently Asked Questions
Can Obsidian read handwriting directly?
No. Obsidian is a markdown text editor with no built-in ink layer and no OCR engine. Every route described above converts handwriting to text or keeps it as an image outside of Obsidian’s own processing; Obsidian only ever sees the result you feed it.
Does reMarkable integrate with Obsidian officially?
No official integration exists. reMarkable’s own ecosystem centers on its Connect subscription for cloud storage and handwriting conversion; getting content into an Obsidian vault requires either that subscription’s export features or unofficial tools like rmapi that talk to reMarkable’s cloud API without any first-party support for the destination being Obsidian specifically.
Is Boox better than reMarkable for a PKM workflow?
For getting text into Obsidian with the least friction, yes — running Android means you install Obsidian directly on the device and avoid a subscription gate for handwriting conversion. For the actual feel of writing, reviewers still generally favor reMarkable’s Canvas display and lower input latency. You’re trading writing feel for pipeline convenience.
Will handwriting OCR ever get good enough for diagrams?
Not with the current generation of engines, which are built around linear text recognition rather than spatial/graphical understanding. Diagrams, mind maps, and free-form sketches are a different problem — closer to image classification than handwriting recognition — and none of the major vendors (MyScript, Boox, Supernote, reMarkable) currently ship a diagram-to-structured-data feature. Treat diagrams as permanent images in your vault, not future text.
Should I pay for Obsidian Sync or use a free alternative for this?
Obsidian Sync isn’t required for any of the handwriting pipelines above — it mainly buys you version history and sim