<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>JayBoard Blog</title>
    <link>https://jayboard.com/blog</link>
    <description>Writing from the JayBoard team — how the app works, how it compares to the other ways of saving what you find, and what we are learning building it.</description>
    <language>en</language>
    <lastBuildDate>Wed, 02 Sep 2026 00:00:00 GMT</lastBuildDate>
    <atom:link href="https://jayboard.com/blog/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Wells — where we save from</title>
      <link>https://jayboard.com/blog/introducing-wells</link>
      <guid isPermaLink="true">https://jayboard.com/blog/introducing-wells</guid>
      <pubDate>Wed, 02 Sep 2026 00:00:00 GMT</pubDate>
      <description>A curated list of 83 sites worth drawing on, filterable by subject and linkable once filtered. The answer to an empty library.</description>
      <content:encoded><![CDATA[<p><img src="https://jayboard.com/media/blog/wells.jpg" alt="The Wells page — a three-column grid of site cards with a filter box and subject buttons above it." width="1760" height="1100" /></p><p>The hardest library to start is an empty one. You install the extension, you understand exactly how it works, and then you sit looking at a grid with nothing in it, trying to remember a single website.</p><p>So we made a list: <a href="https://jayboard.com/wells">Wells</a>, 83 places worth drawing on.</p><h2 id="what-is-on-it">What is on it<a class="hash" href="#what-is-on-it" aria-label="Link to this section">#</a></h2><p>Space archives that release everything at full resolution. Museums with real open-access programmes — the Met, the Rijksmuseum, the Getty, the Smithsonian. Natural history plates scanned out of century-old books. Game art, asset libraries and fifteen thousand screens of game interface. Type specimens written like essays. The feeds where visual work turns up before it is anywhere else.</p><p>Every entry is a site we save out of ourselves. Nobody paid to be on it, and there is no affiliate link anywhere on the page — if that ever changes, the page will say so.</p><h2 id="filtering-it">Filtering it<a class="hash" href="#filtering-it" aria-label="Link to this section">#</a></h2><p>The box at the top filters as you type across the name, the address, the subject and the description. Every word has to match, so two words narrow rather than widen: <code>game-dev ui</code> gives you the sites that are both.</p><p>Under it are the sixteen subjects that recur most across the list — not a list we typed out, but the tags the entries actually carry, counted. When the list changes, so do the buttons.</p><h2 id="tell-us-what-is-missing">Tell us what is missing<a class="hash" href="#tell-us-what-is-missing" aria-label="Link to this section">#</a></h2><p>It is curated by hand, which is the only reason it is short enough to read, and it is certainly missing things you know about. Send them over.</p>]]></content:encoded>
    </item>
    <item>
      <title>JayBoard vs Karakeep</title>
      <link>https://jayboard.com/blog/jayboard-vs-karakeep</link>
      <guid isPermaLink="true">https://jayboard.com/blog/jayboard-vs-karakeep</guid>
      <pubDate>Mon, 31 Aug 2026 00:00:00 GMT</pubDate>
      <description>Both save what you find on the web, and they are not the same tool. One keeps a library you read; the other keeps one you look at.</description>
      <content:encoded><![CDATA[<p><img src="https://jayboard.com/media/board-dark.jpg" alt="A JayBoard moodboard, with saved images arranged as cards on an infinite canvas." width="1760" height="1100" /></p><p>People who try JayBoard have often tried <a href="https://karakeep.app" target="_blank" rel="noopener">Karakeep</a> first, and the question that follows is fair: they both save things off the web, so what is the difference?</p><p>The short answer is what happens <em>after</em> the save. Karakeep gives you a library you read. JayBoard gives you one you look at.</p><h2 id="what-karakeep-is">What Karakeep is<a class="hash" href="#what-karakeep-is" aria-label="Link to this section">#</a></h2><p>Karakeep — formerly Hoarder — calls itself "a self-hostable bookmark-everything app with a touch of AI for the data hoarders out there". It saves links, notes, images and PDFs, archives the page so the bookmark outlives the site, tags everything with an LLM, and puts it all behind a very good search. It is open source, it runs on your own machine with Docker, and it is built around reading: a reader view, highlights, feeds, a phone in your pocket.</p><h2 id="what-jayboard-is">What JayBoard is<a class="hash" href="#what-jayboard-is" aria-label="Link to this section">#</a></h2><p>JayBoard is a private visual library. What you keep in it is media and bookmarks — images, videos, links and whole pages saved offline — and the assumption is that you will want to look at them again rather than work through them as a queue. It is built to be quick at every step of that: the capture, what gets stored, how fast the grid comes back, and what you can arrange out of it afterwards.</p><p>That assumption changes the shape of the app. Capture is an overlay you aim at a page: <kbd>Alt</kbd><kbd>Shift</kbd><kbd>S</kbd> to pick several images at once, <kbd>Alt</kbd><kbd>Shift</kbd><kbd>D</kbd> to send whatever you click straight to a project with no form in the way. Clicking five images on a page should not be five forms, and in JayBoard it is five clicks.</p><p>The AI pass then writes tags and a description <em>of the thing itself</em> — what the image shows, what the video is, what the page says — so a screenshot of a dark navigation bar comes back for the words you would actually think of a month later, not for whatever the page it came from happened to be called.</p><p>And the library has a second half that a bookmark manager has no reason to have: infinite moodboards, where the things you saved become cards you can drag around, group into frames, annotate and connect, then export as PNG or SVG. That is the dividing line. Lists and tags are the right structure for a queue of things to read. For a pile of references they are not enough — the pile only becomes useful once you can lay it out next to itself.</p><h2 id="side-by-side">Side by side<a class="hash" href="#side-by-side" aria-label="Link to this section">#</a></h2><div class="tablewrap"><table><thead><tr><th></th><th>JayBoard</th><th>Karakeep</th></tr></thead><tbody><tr><td><strong>Picking several things off one page</strong></td><td>✅ Aim the overlay at images and videos, or one click each</td><td>❌ A page at a time</td></tr><tr><td><strong>An infinite canvas to arrange them on</strong></td><td>✅ Cards, notes, frames and connectors, exported as PNG or SVG</td><td>❌ Lists and tags</td></tr><tr><td><strong>AI tagging</strong></td><td>✅ Tags and a description written for what was saved — the image, the video, the page — at four levels of detail</td><td>✅ Tags and summaries, and it will use a local Ollama model</td></tr><tr><td><strong>Search</strong></td><td>✅ Full text over titles, tags, descriptions, your notes, URLs, filenames, and the text of pages saved offline</td><td>✅ Full text over everything it stores, plus semantic search and OCR inside images</td></tr><tr><td><strong>Saves video</strong></td><td>✅ With yt-dlp</td><td>✅ With yt-dlp</td></tr><tr><td><strong>Whole-page archive</strong></td><td>✅ The whole page, saved as one self-contained HTML file</td><td>✅ The whole page, saved as one self-contained HTML file</td></tr><tr><td><strong>Sharing</strong></td><td>✅ A read-only link to a board</td><td>✅ Lists, with editors and viewers</td></tr><tr><td><strong>On a phone</strong></td><td>❌ The library is a web app; capture is desktop Chrome</td><td>✅ iOS and Android apps</td></tr><tr><td><strong>Self-hosting</strong></td><td>❌ Hosted by us</td><td>✅ Docker, free forever</td></tr></tbody></table></div><h2 id="which-one">Which one<a class="hash" href="#which-one" aria-label="Link to this section">#</a></h2><p>Take Karakeep if you want to run it yourself, on your own disk under your own domain — JayBoard is hosted, and if that is the deciding factor there is nothing more to weigh. Take it too if you read on a phone, or if most of what you keep is prose.</p><p>Take JayBoard if what fills your saves is images and video, if you want them on a board rather than in a list, and if the part that has beaten you before is the saving itself.</p><p><em>Everything above about Karakeep was checked against its own site and README when this was written. It is an actively developed project and this will go out of date; if we have got something wrong, tell us and we will fix it.</em></p>]]></content:encoded>
    </item>
    <item>
      <title>Introducing JayBoard</title>
      <link>https://jayboard.com/blog/introducing-jayboard</link>
      <guid isPermaLink="true">https://jayboard.com/blog/introducing-jayboard</guid>
      <pubDate>Sun, 23 Aug 2026 00:00:00 GMT</pubDate>
      <description>A private visual library for the images, videos and pages you find on the web — captured in a click, described by AI, and arranged on a canvas.</description>
      <content:encoded><![CDATA[<p><img src="https://jayboard.com/media/gallery-dark.jpg" alt="The JayBoard library, a grid of saved images and pages with projects listed down the left." width="1760" height="1100" /></p><p>JayBoard is live at <a href="https://jayboard.com" target="_blank" rel="noopener">jayboard.com</a>. This is what it is and why it exists.</p><h2 id="the-thing-that-keeps-happening">The thing that keeps happening<a class="hash" href="#the-thing-that-keeps-happening" aria-label="Link to this section">#</a></h2><p>You find a picture worth keeping — a layout, a colour, a piece of type, a frame from a video — and you save it. Into a bookmark folder you will never open, a tab you will lose, a screenshot on the desktop named <code>Screenshot 2026-08-19 at 14.03.11.png</code>. Six weeks later you know you have seen the thing. You cannot find it.</p><p>The tools that handle this well handle <em>articles</em>. They save the words, make them searchable, and give you them back to read. That is the right design for a queue of things to read, and it is the wrong one for a pile of pictures, because the thing you remember about a picture is never the words on the page it came from.</p><p>JayBoard is built for the pile of pictures.</p><h2 id="saving-is-a-click">Saving is a click<a class="hash" href="#saving-is-a-click" aria-label="Link to this section">#</a></h2><p>Capture is an overlay you aim at the page. <kbd>Alt</kbd><kbd>Shift</kbd><kbd>S</kbd> brings it up, you click the images you want, and one form at the end says where they go. Or <kbd>Alt</kbd><kbd>Shift</kbd><kbd>D</kbd>, where each click sends what is under it straight to your active project with no form at all.</p><p><img src="https://jayboard.com/media/changelog/one-click-capture.jpg" alt="The one-click overlay outlining a picture on a page, with a toolbar reading &quot;Sends to Inbox&quot;" width="1760" height="1100" loading="lazy" /></p><p>That is the part that has to be cheap. Clicking five images on a page should not be five dialogs, and if saving costs you a decision every time, you stop saving.</p><p>The extension also takes a video, a link, or a whole page — saved as one self-contained file, so it still opens after the site behind it is gone.</p><h2 id="then-it-describes-what-it-saved">Then it describes what it saved<a class="hash" href="#then-it-describes-what-it-saved" aria-label="Link to this section">#</a></h2><p>Everything that arrives can be annotated: tags, a description, or both, written from the picture rather than from the page it came from. A screenshot of a dark navigation bar gets described as a dark navigation bar, so it comes back a month later for the words you would actually think of.</p><p><img src="https://jayboard.com/media/changelog/annotate.jpg" alt="The library with the Annotate menu open, offering tags, a description, or both" width="1760" height="1100" loading="lazy" /></p><p>Search then runs in full text across titles, tags, those descriptions, your own notes, addresses, filenames, and the whole text of any page you saved offline.</p><h2 id="and-then-you-can-lay-it-out">And then you can lay it out<a class="hash" href="#and-then-you-can-lay-it-out" aria-label="Link to this section">#</a></h2><p>The half that makes it a visual library rather than a tidy folder: every board is an infinite canvas. Drag things out of the library onto it, group them into frames, write on them, connect them, and export the result as PNG or SVG.</p><p><img src="https://jayboard.com/media/board-dark.jpg" alt="A moodboard: saved images arranged as cards on an infinite canvas" width="1760" height="1100" loading="lazy" /></p><p>A pile of references is not useful until you can see it next to itself. That is what a canvas is for, and it is why a list of links was never going to be enough.</p><h2 id="what-it-is-not">What it is not<a class="hash" href="#what-it-is-not" aria-label="Link to this section">#</a></h2><p>There is no feed. There is no algorithm, nothing recommended, and nobody else's material in your library. Every account gets its own store, and that separation is structural rather than a filter we remember to apply.</p><p>It is a hosted product with a Chrome extension, which means there is nothing to run and nothing to maintain — and it means it is not self-hosted. If that is a dealbreaker, it is an honest one.</p><h2 id="try-it">Try it<a class="hash" href="#try-it" aria-label="Link to this section">#</a></h2><p>There is a <a href="https://app.jayboard.com/api/demo" target="_blank" rel="noopener">live demo</a> that needs no account: a read-only library with a moodboard in it, which is the fastest way to tell whether this is the thing you have been missing. The <a href="https://jayboard.com/docs">documentation</a> covers the rest, and the <a href="https://jayboard.com/changelog">changelog</a> is where everything new shows up.</p>]]></content:encoded>
    </item>
  </channel>
</rss>
