--- title: "Bring your own sources to Claude: how Kepler makes Claude even more powerful" url: "https://www.keplerbrowser.com/blog/bring-your-own-sources-to-claude" description: "Claude is a brilliant reasoning engine — but it works from the sources it can reach, not the ones you care about. Here's how Kepler gives Claude a portable memory layer of your own saved articles, videos, and posts, connected in minutes over MCP." --- Claude is a brilliant reasoning engine. Kepler is the information tool that lets you feed it _your_ sources — the articles, videos, and posts you’ve saved — so Claude works from what you actually care about. Because Claude speaks MCP natively, connecting a Kepler space takes minutes. Every AI model works from the sources _it_ decides to reach. That’s fine for general questions. But the best answers come from _your_ material — the five articles you read this week, the TikTok of a lunch spot you want to try, the reel of somewhere you’re planning to travel. Kepler lets you collect those sources yourself and hand them to Claude, turning a general reasoning engine into one that reasons over your digital world. **Kepler is an information layer for Claude.** It’s not another AI model and it doesn’t replace Claude — it captures the things you find useful, keeps them as portable memory across every device, and makes them available to Claude when you need them. The result is a more personalised Claude: same model, grounded in your context. > Kepler plugs into Claude through MCP — the open standard Anthropic created for connecting AI to outside context. Claude was built to speak it, so Kepler connects as a first-class connector, not a workaround. ## Your sources, not just the model’s A model on its own is limited to what it can see: its training data, whatever it can search, and the handful of things you paste into a chat. Kepler flips that. _You_ decide what goes in — save an article, a paper, a YouTube video, a post — and that curated set becomes context Claude can draw on. It’s the difference between asking a smart stranger and asking someone who’s read everything you’ve read. - **You choose the inputs.** Kepler only sees what you save. Curating your own sources gives Claude a sharper, more relevant starting point than any general search. - **Annotations add context.** Add your own annotations to a source and they become extra context Claude can utilise — telling it not just _what_ you saved but _what mattered_ in it, in your words. - **It gets better the more you save.** Your saved space is an asset that compounds: every save makes Claude’s answers on that topic more grounded. - **Build sources together.** Spaces can be collaborative — you and a teammate, partner, or friend can add to the same space or share sources with each other, so Claude draws on what you’ve _all_ saved, not just you. ## Capture anywhere, remember everywhere Kepler is URL-first and built for how you actually come across things. Read something great on your laptop, catch a TikTok on your phone, find a paper on a tablet — save it into a **space** (think: a project folder for context) from the web extension or the mobile share sheet. It’s instantly part of a portable memory system that travels with you across every device. So when you sit down with Claude — on any device — the context is already there. Nothing to re-find, nothing to re-paste. ## Built for how power users work with Claude Kepler is designed to capture and share information _across_ your Claude workflow, not just a single chat: - **Across chats.** Start fresh conversations without starting from zero — your saved space is available in every one. - **Across agents.** Point Claude agents at the same curated context so they work from your sources, not just the open web. - **Across people.** Share a space with a teammate and you share the memory behind it — everyone’s Claude draws on the same curated set. ## Step by step 1. **Create a space** for what you’re working on — “Q3 competitor research,” “Japan trip,” “thesis sources.” 2. **Save as you browse.** Drop links into the space from the web extension or mobile share sheet. Highlight the lines that matter. 3. **Let Kepler pull context.** Each save gets a basic extraction automatically (title, summary, text). For dense sources — a long PDF, a video with no transcript — trigger a deep extract. 4. **Connect Claude to your space** via the Kepler connection — full walkthrough in [Connect Kepler to Claude](https://www.keplerbrowser.com/help/articles/5169157-connect-kepler-to-claude). 5. **Ask grounded questions.** Claude now reasons over the sources you chose. ## What this unlocks **Research synthesis** > “Summarise the main arguments across the RAG-eval articles in my Research space, and flag where they disagree.” → _(Claude pulls the five saved sources)_ “Across the five, three argue retrieval quality dominates; two argue annotation/curation matters more. They disagree most on…” **Trip planning, together** > “Build a 7-day itinerary from the reels and posts my partner and I saved in our Japan space.” → _(Claude reads everything you both saved)_ “Here’s a route built around the spots you both flagged, with the two restaurants from the reels you saved slotted in…” **Picking up weeks later** > “Remind me what I concluded in my thesis-sources space and what’s still unresolved.” → _(Claude reads the annotated set)_ “Your saved sources converge on X; the open question you flagged twice is Y…” Same model, more personal answers — because it’s finally working from your material. ## Why not just paste? Pasting doesn’t scale, loses the source, and disappears when the chat ends. Kepler does the work pasting can’t: a clean UI to organise your sources, automatic extraction and embedding of everything you save, and retrieval that’s optimised so Claude pulls the _right_ sources for the question. A Kepler space is **portable** across devices, **reusable** across chats and agents, and **yours** — a curated, annotated set that gets more useful every time you save. ## Get started **Connect Kepler today — it’s free.** Install the extension, start a space, and connect it to Claude — the step-by-step guide is in [Connect Kepler to Claude](https://www.keplerbrowser.com/help/articles/5169157-connect-kepler-to-claude). [Get Kepler](https://www.keplerbrowser.com/get-kepler) to start. Connecting uses MCP, which is **free and unlimited on every tier** — it never costs AI credits. > Kepler helps you create context for AI from the links you save. It improves the AI tools you already use — it isn’t one itself.