AI knowledge graph · personal knowledge management
Hop through everything you know.
Memohopper turns pages, PDFs, emails, tasks, and quick notes into connected knowledge pages on a visual graph — for researchers, builders, and anyone whose ideas outgrow a folder of notes. AI keeps it sourced and searchable. You do the exploring.
Free plan included · No credit card · Your agents connect free
app.memohopper.com · Launch workspace
What is Memohopper?
Memohopper is an AI knowledge tool that turns everything you read, clip, and ask into connected knowledge pages on a visual graph. You explore it by hopping node to node; AI keeps it sourced, searchable, and growing — and your AI agents can use the same knowledge over MCP, on every plan.
Reason mode: 53 → 93 on our multi-hop retrieval eval10+ capture sources from clipper to connectorsAgents free on every plan — read and write
Explore
Type a concept. Get a page. Watch it link in.
Exploring is your move. Type any concept and Memohopper generates a full, sourced Context — a readable knowledge page — then links it into your graph next to what you already know.
A complete, readable page — not a list of search results
Sources attached; concepts inside the text link onward
New nodes join the graph where they belong
You decide what to explore next — the graph never grows without you
Graph · Launch workspace
Hopping
Read your way across the graph, node by node.
Click a node, read, follow a typed edge — depends_on, uses, derived_from — to the next idea. Colors tell you what kind of thing you're looking at; edge labels tell you why it's connected.
Typed, colored nodes and labeled edges
Breadcrumbs remember the path you took
Temporary (dashed) nodes you promote when they prove useful
One chat, scoped by a chip — this Context ⇄ whole workspace. Answers stream in with citation chips that point at real nodes, memos, and web sources. Nothing enters the graph unless you harvest it.
One box, routed by intent — ask, edit, search the web, or “Think deeper”
Citations you can click — to your own pages, not vague confidence
Drop files straight into chat
Keep a good answer: harvest it as a memo or a new Context
Chat · whole workspace
AI edits
AI that edits the page, not the thread.
Ask for a change and you get an editable proposal card — never a silent rewrite. Review the diff, then choose [Rewrite in place] or [Add as a block]. The page gets better; the answer doesn't scroll away.
Proposal cards with a viewable diff
Rewrite in place, add as a block — or just answer instead
Every improvement lands on the page, where you'll find it again
Edit proposal
Reason mode
Answers a single search can't reach.
Some answers hide behind another fact. Reason mode reads what it finds, names the missing bridge, searches again — and shows you the hop trail. In our internal multi-hop retrieval eval it nearly doubled accuracy vs single-shot retrieval (53 → 93).
Multi-hop retrieval with visible steps
Included on the free plan
On Pro, chat auto-escalates to Reason mode when a question needs it
missing bridge: "Mem's gap" — searched again found: competitor landscape
Churn concentrates at the pricing step: review-mining shows users leave when AI feels metered Beta feedback themes, and Mem's gap is exactly there — no agent access on free Competitor landscape.
Capture · 10+ integrations
Every entry point builds Contexts.
The New-entry launcher has three tabs — Prompt (ask, and the AI builds a Context), Document (upload & extract), Write (author one yourself). On paid plans, connectors bring in the rest from where it already lives.
Supported integrations
Web clipper, email-in, URL scraper, custom webhook
Google Drive, Google Docs, Gmail, Google Calendar, Todoist, Jira, Confluence, ClickUp
File upload with vision OCR for scans
Notion, Linear, Dropbox, OneDrive & more landing soon
New entry · Prompt / Document / Write
Settings · Connections
Memos & Library · verbatim capture
The memo in Memohopper.
Some things should stay exactly as you wrote them. Memos are verbatim amber sticky-notes that live on the graph itself; the Library keeps every capture accounted for, with its origin on the label.
Verbatim memos — never paraphrased by AI
Library views: All · Inbox · Forgotten
Provenance badges: web clipper, email-in, agents, connectors
Hybrid semantic + keyword search across everything
Library · All / Inbox / Forgotten
For agents
A shared brain for your AI agents.
Memohopper ships an MCP server and a REST API with scoped tokens. Your agents read and write the very same graph you explore — and full agent access is free, on every plan. What an agent stores this week, you can hop through next week.
MCP server + REST with scoped tokens
Read and write — free on every plan
One shared memory across all your agents — and you
// your agent's toolbox, over MCPremember("Beta users stall at the pricing step")
→ stored · linked toPricing ladderdeep_recall("what did we decide about free agents?")
→ 3 contexts ·"the plan governs capability,
not the channel"· fromagents-free wedgeevolve_context("billing-via-paddle", "add the
verification-timeline caveat")
→ proposal created — you review the diff
Dashboard · activity
Plain words
The hard parts, explained simply.
What's a Context?
A Context is a living page for one idea — readable, editable, with its sources attached. When AI improves it, the page itself gets better, in place. Compare that to a chat answer, which scrolls away the moment you ask the next question.
Why a graph?
Knowledge is mostly connections — this idea depends on that one, this decision came from that evidence. A graph makes those connections visible and walkable, so instead of remembering where you filed something, you follow the lines between ideas.
What's Reason mode?
A way of answering questions that need two steps. It searches, reads what it finds, realizes what's still missing, and searches again for that — then shows you each step it took, so you can check the trail, not just trust the answer.
How it works
Capture once. Keep forever. Find it thinking.
01
Capture
Clip a page, forward an email, upload a PDF, connect a source — or just write. Every entry point becomes a Context on your graph.
02
Hop & Read
Open the graph, click a node, read the page, follow a typed edge to the next idea. Breadcrumbs keep your trail.
03
Explore & Evolve
Ask for what's missing and AI builds the page. Ask for changes and it proposes an edit in place — you approve the diff.
04
Recall
Search, chat, or Reason mode — answers grounded in your own content, with citations. Your agents can do the same.
Use cases
What people use Memohopper for
Research
Capture papers, PDFs, and web sources; each becomes a sourced Context linked to related work. Reason mode answers multi-hop questions across your literature — with citations back to the sources.
Product & decision work
Clip competitor pages, forward customer emails, connect Jira, Todoist, or Confluence. Decisions stay linked to the evidence they came from — walkable when someone asks “why?” months later.
Memory for AI agents
Give your agents a persistent, shared memory over MCP — free on every plan. What one agent stores, another can recall, and you can read it all on the same graph.
Compare
Memohopper vs a notes app vs an AI chatbot
An honest comparison of where each one puts your knowledge — and what happens to it afterwards.
Memohopper
A notes app
An AI chatbot
Where knowledge lives
Connected, editable pages on a visual graph
Documents in folders
Chat scrollback
Connections between ideas
Typed, labeled edges you can see and walk
Manual links, if you make them
None are kept
AI answers grounded in your content
Yes — with citation chips to your own pages
Sometimes, as a paid add-on
Grounded in the web, not your notes
AI agents can read / write it
Yes — MCP + REST, free on every plan
Rarely, and usually read-only
No shared, persistent store
What happens to a good answer
Becomes a page, linked into the graph
You paste it somewhere yourself
It scrolls away
Privacy & security
Your knowledge stays yours.
Your content is never used to train AI models, never sold, and never shown ads. AI features send only the relevant content to model providers to generate your response — nothing more. Full details in the privacy policy.
No training on your content, everEU-hosted — Supabase (eu-west-1) + Fly.io AmsterdamNo ads, no data salesNo tracking cookies on this site
Pricing
Start free. Upgrade when your graph does.
Every plan includes the full toolkit — every AI feature, even Reason mode, and free agent access. Plans differ mostly in capacity; paid plans add premium models, integrations, and background intelligence.
No daily lockouts. Your AI allowance refills continuously — no midnight reset.
Prices in USD. Taxes handled at checkout. Payments processed by Paddle · Refund policy
FAQ
Questions, answered first.
What is Memohopper?
Memohopper is an AI knowledge tool that turns everything you read, clip, and ask into connected, editable knowledge pages — called Contexts — on a visual graph. You explore the graph by hopping from node to node, AI keeps every page sourced and searchable, and your AI agents can read and write the same knowledge over MCP.
Is there a real free plan?
Yes — the Free plan is $0 forever and includes the whole product with a generous AI allowance. Every AI feature is included, even Reason mode; agents connect free with read and write access; and you get 2 workspaces, 500 Contexts, capture by prompt, document, or write, plus hybrid search and the Library. No credit card required.
What is a Context?
A Context is a structured, readable, editable knowledge page that Memohopper creates from anything you capture — a web page, PDF, email, task, or quick note. It lives as a node on your graph, linked to related ideas, and AI can improve it in place instead of replying in a thread that scrolls away.
What is Reason mode?
Reason mode is Memohopper's multi-hop retrieval for questions a single search can't reach: it searches, reads what it finds, names the missing bridge, searches again — and shows the hop trail behind its answer. In our internal multi-hop retrieval eval it nearly doubled accuracy compared with single-shot retrieval (53 → 93).
Can my AI agents use Memohopper?
Yes — Memohopper ships an MCP server and a REST API with scoped tokens, and agent access is free on every plan, for reading and writing. Your agents use tools like remember, recall, deep_recall, and explore to work with the very same graph you see in the app.
How is Memohopper different from Notion or Obsidian?
Notes apps store documents in folders and leave the linking to you; Memohopper turns what you capture into connected knowledge pages on a graph, with typed, visible connections you can walk. Chat answers cite your own content, AI improves pages in place, and your AI agents can read and write the same knowledge — free on every plan.
How is Memohopper different from ChatGPT?
A chatbot's answers scroll away; Memohopper turns good answers into permanent, editable pages linked into your graph. Its chat is grounded in your own content with citations to real nodes and memos, and everything it produces stays organized, searchable, and reusable — by you and by your agents.
Is my data private, and is it used to train AI?
Your content is never used to train AI models, is never sold, and there is no advertising. Your data is hosted in the EU; AI features send only the relevant content to model providers to generate your response; and this site sets no tracking cookies. See the privacy policy for full details.
How do I cancel or get a refund?
You can cancel any time from the app's billing settings or via the link in any Paddle receipt, and you keep paid features until the end of the period you've paid for. Your first purchase has a 14-day money-back guarantee — email support@memohopper.com. Full details are in the refund policy.
Where is my data stored?
In the European Union: the database and storage run on Supabase (eu-west-1) and the application is hosted on Fly.io in Amsterdam. Where a subprocessor is outside the EU, transfers rely on Standard Contractual Clauses or an adequacy decision — the full list is in the privacy policy.