Sold by capacity, not by tokens.
Every plan ships the full product — OCR, citations, semantic search, MCP endpoint. The only thing that changes between tiers is how much you can keep on the shelf at once.
§ The plans
Three shelf sizes. Same room.
Limits are enforced exactly as listed. No asterisks, no roadmap boxes, no placeholder checkmarks.
§ In every plan
Full product on every shelf.
Pricing gates capacity, not capability. Nothing below is locked behind a higher tier.
OCR
Full OCR with page-accurate citations
Every PDF is processed page by page. When you — or Claude — ask a question, the answer cites the exact page it came from.
Search
Full-text and semantic search, side by side
Search is scoped per knowledge base, so unrelated material never leaks in. Both keyword and vector search are included on every plan.
MCP-native
A personal wiki endpoint for Claude, Cursor, and friends
Expose your library to Claude.ai, Cursor, or any MCP client through a personal API key. The wiki becomes a tool your agents can actually use.
Your data
Not used to train models
Your documents are stored for you, indexed for you, and served back to you (and to MCP clients you personally authorise). Nothing else.
§ Why upgrade
When Free stops being enough.
We built Free to be usefully generous, not a teaser. Upgrade when the shape of your work has changed — not because a countdown expired.
Your research stops fitting in a chat window.
Free is generous for a project or a course. Once you are juggling several threads at once — or reading across a hundred PDFs — you want Pro’s 3,000 pages and 10 knowledge bases to keep them properly separated.
Claude actually uses the wiki you built.
More API keys means you can wire the MCP endpoint into Claude.ai, Cursor, custom agents, and a scratch account for experiments — without rotating keys or stepping on yourself.
The archive outlives the project.
Capacity, not tokens. Nothing resets at the end of the month. Pro and Max exist so that a year of careful reading is still there, fully searchable, when you come back to write the thing up.
It is the total number of OCR-processed pages your account holds at once — a capacity limit on your library, not a monthly quota that resets. Delete a document and its pages free up again.
Your existing content keeps working. Reading, search, and MCP access are unaffected. New uploads or new knowledge bases are paused until you either free space or move to a larger plan. You are never auto-billed for overages.
Upgrading to a paid plan opens a hosted Creem checkout. Any change after that — switching tiers, updating payment details, or cancelling — happens in the Creem customer portal, which you can reach from the manage button on this page once you are subscribed.
Your subscription enters a short grace window so service continues while the issue is sorted out in the Creem portal. If it is not resolved by the end of the period, the account drops to Free and any content above Free limits becomes read-only until you trim it down or resubscribe.
No. Your sources and wiki pages are stored in your account and used only to serve your own conversations and your own MCP clients. They are not used to train any model.
Not yet. Today we only offer monthly Free, Pro, and Max for a single user. If you need seats, invoicing, or annual billing, email [email protected] and we will keep a list.
Visa, Mastercard, American Express, PayPal, Apple Pay, Google Pay, SEPA, and local payment methods surfaced based on your region. Checkout is hosted, so the exact set you see on the final step depends on where you are buying from.
§ One last thing
The documents you already have deserve a better shelf.
Start on Free, put a handful of real documents through, and see if the citations hold up under your usual questions. Upgrade only when the shelf starts feeling tight.