Contextick FAQ

Because most AI chats start fresh and don’t carry the specifics of what came before — ChatGPT isn’t unusual here. Why that happens, and what people do about it, is its own post.

Built-in AI memory usually works by summarizing — it keeps the gist and lets the exact details go. Contextick keeps the exact text you saved and hands it back word for word, so the specifics survive.

Only partly, and only inside that one AI. Built-in memory is summarized, and it decides what’s worth holding on to, so you don’t get to see all of it. Contextick keeps the full text you saved and works across every AI you connect.

Because there’s usually nowhere else to put what your AI should remember, so you file it yourself — in a doc, a note app, or a folder — and paste it back in later. Contextick removes that step: you save once through your AI, and it pulls the right piece back when you ask.

Contextick is a private knowledge layer for your AI. You save any text once, and any AI you connect can recall it by meaning in later conversations.

No. There’s no app to open and nothing to organize — you save through the AI you already use, and your AI does the recalling. It’s a store your assistant reads from, not another place for you to file things.

In spirit, yes — but you don’t build it. A second brain is a system you assemble; Contextick is just a store your AI reads from. There’s nothing to set up beyond connecting it.

No app or extension. You connect Contextick to an AI you already have — like Claude or ChatGPT — and start saving.

It means a separate store your AI can read from and write to. Your knowledge lives in that layer instead of inside any one assistant, so it stays available no matter which AI you’re using.

Any text — notes, decisions, documents, drafts, code, or facts you want your AI to remember. If it can be written as text, you can save it.

Yes, and there’s no minimum length. That said, a few lines with real context will serve you better later than a one-liner — and since every saved note takes up a similar minimum amount of space, a single sentence costs about as much room as a short paragraph. Save the thought, not just the fragment.

Yes. A long document fits in a single note; very long ones are better split into a few parts, each saved on its own.

Yes — two ways. Ask your AI to read the file and save the text, or upload the file yourself from your dashboard and it is converted and saved for you.

Saving is instant, but an AI can only find a document once background processing finishes. That usually takes a few minutes, and up to 24 hours when the queue is busy. You can open and read it in your list the whole time, and we email you when it is done.

Your list marks it Not searchable: the text is saved and you can still read it, but no AI will find it. A long document is stored as several items, so this can affect only part of one. Open the item to retry the parts that failed — you never need to upload the file again. If the rest of the document is still processing, the retry appears once that finishes.

Yes. Meeting summaries, research notes, blog drafts, and similar text are exactly what Contextick is for — save them once and any AI can build on them later.

Yes. Conventions, snippets, architecture notes, and API details are all text, so they save and come back like anything else — useful across ChatGPT, Claude, Cursor, and other coding tools.

Yes — and you don’t need ChatGPT’s export for it. Once Contextick is connected, open an older conversation and ask ChatGPT to summarize it and save it. The same works in Claude or any connected AI, as long as the assistant can still read that thread. You’re not limited to what you say from today on.

You can, but it isn’t the best way to use it. A long conversation gets split across several notes, and it comes back in pieces when you look for it later. Ask for the conclusion instead — “summarize what we’ve worked out and save it to Contextick” — and you get one tidy note that’s easy to find again.

Not directly — Contextick stays focused on text. Keep media where it already lives and save the link, so your AI can point you back to it.

No. Contextick has no folders — it indexes what you save by meaning, so things are found by what they say instead of where you put them.

No tags, no categories. Contextick works from what your notes mean, so there’s nothing to label.

No. A short summary helps, but you don’t need to invent titles or keywords — a rough sense of what you saved is enough for Contextick to find it later.

Yes — that’s the point. You don’t track where anything is; you just ask your AI, and Contextick finds the closest matches by meaning.

Because Contextick does the part organizing was for. When you save, it indexes your text by meaning, so it can be found later without any structure you have to build and maintain.

You ask; Contextick finds. It matches your question against everything you’ve saved and returns the closest items by meaning — no exact keywords needed — and your AI answers from them. The exact text you saved is there when you need the full detail. The docs show the tools your AI uses to do this.

Usually you ask, and your AI queries Contextick to answer. It can also bring in something relevant on its own when it fits the conversation.

Yes. Contextick indexes your notes by meaning rather than by wording, so a note about “our refund policy” can surface when you ask about “returns,” even without matching words.

Contextick compares the meaning of your question with the meaning of what you’ve saved and returns the closest matches. You don’t set rules — the closest-in-meaning notes come back.

Yes. Contextick can return several related notes, and your AI combines them into one answer — so scattered pieces come back together.

Yes — that’s a fair description. RAG means retrieving the relevant text and letting an AI generate an answer from it. Contextick is the retrieval half, over your own knowledge: it indexes what you save so it can be found by meaning (semantic search), and your AI does the generating. You don’t need the jargon to use it — you save, you ask.

ChatGPT’s memory works only inside ChatGPT and is summarized. Contextick keeps the exact text and works across every AI you connect, so the same knowledge follows you when you switch.

Claude Projects keep context inside Claude. Contextick isn’t tied to one assistant — save something with Claude and recall it in ChatGPT or a coding tool later.

They’re great when you want to organize knowledge by hand — folders, tags, links. Contextick is built for people who’d rather not organize at all: you just save, and Contextick indexes it by meaning so your AI can bring it back.

The upkeep. Second brain apps ask you to file, link, and prune so you can find things later; Contextick indexes by meaning, so there’s no system to maintain. If you enjoy tending one, keep it.

Pasting works until your notes outgrow a single message or you can’t find the right one. Contextick holds far more than a chat can, and your AI pulls only the relevant parts when you ask.

Any AI that supports MCP (the Model Context Protocol) — including ChatGPT, Claude, Cursor, Devin (formerly Windsurf), and Replit. You connect once per AI and your saved knowledge is available in each. We have step-by-step connection guides for each of them.

Yes — that’s the core idea. Save something in one AI and recall it in another, because your knowledge lives in Contextick, not inside any single assistant. Contextick also records which AI each note came from, so you can tell what you saved in ChatGPT from what you saved in Claude.

Search for Contextick in Claude’s connector directory and connect it — Settings → Connectors in Claude. It works on any Claude plan, the free one included, and you sign in with Google rather than pasting a key. Step-by-step guide for Claude.

Open Settings → Plugins, search for Contextick, and install it from ChatGPT’s plugin directory. It works on any ChatGPT plan, the free one included — no Plus subscription is required. Step-by-step guide for ChatGPT.

In Cursor or Devin (formerly Windsurf), add Contextick to your MCP config with an API key. In Replit, add it from the MCP settings pane in two clicks — no config file needed.

They work the same way: add Contextick to that tool’s MCP config. Claude Code and Codex CLI both speak standard MCP, so what you saved in a chat app is available from the terminal too — and the other way around. Step-by-step guide for command-line tools.

Any AI that supports MCP. MCP is an open standard, so support isn’t something we add one assistant at a time — Contextick is a standard MCP server, which means any AI or tool that can add your own MCP server works with it today. The list above is just where we’ve written out the steps.

Contextick itself is free to use. Claude, ChatGPT, and Cursor all connect on their free tiers. A few hosts gate MCP behind their own paid tier — Replit's MCP support is part of its paid Agent, for example — but that's the AI's pricing, not ours.

Not in the Gemini web app today — it has no self-serve way to add your own MCP server, so there’s nothing for you to add Contextick to. Google’s Gemini surfaces are changing quickly, and we’ll announce it here when the app supports it.

You do. Your knowledge lives in your own private store and stays yours.

In a private database dedicated to your account.

No. Contextick stores and retrieves your knowledge; it does not train models on what you save.

Yes, anytime. You can delete any saved item, and deleting it frees up your space right away.

Yes. From your account page you can download everything you’ve saved as a single file — one line per item, with the full text, so it opens in anything. It’s the thing to grab before you delete your account.

Only what you ask for. Your AI reads a note when it’s relevant to your question; your store isn’t handed over wholesale, and it’s never sold.

Only what you’d want your AI to reuse. Your store is private and never used to train models, but a good habit is to keep secrets out of it — passwords, card numbers, and the like. If something sensitive is mixed into text you want to save, ask your AI to mask those details before saving.

Yes. The Free plan includes 5 MB of storage at no cost.

The free 5 MB holds a few full-length books’ worth of text — hundreds of pages. How much fits depends on how long your notes are.

It varies, but pulling things back out is one of the most common meters. Free tiers get capped by monthly lookups, by credits, or by how much text gets processed — and many meter saving as well, so what you’re charged for is activity, not the amount you’re keeping. The ones that meter operations usually run an AI pass of their own when you save, so counting operations matches what it costs them. Contextick skips that pass: the assistant you’re already talking to writes the title and summary, so capacity is the only thing that counts against your plan.

Yes — about 16,000 characters per saved note (roughly 2,500–3,000 words). For longer material, save it in parts.

Yes. Pro is $10 per month for 1 GB of storage — 205× the Free plan. You can upgrade from your account page.

Your plan stays active until the end of the current billing period. Whenever a paid plan ends — you cancel it, you take a refund, or a payment can’t be recovered — if you’re over the Free limit at that point everything you’ve saved is removed (not just the amount above the limit) — so keep your own copy of anything you want to hold on to before then. Resubscribing before that happens keeps everything. On the Free plan, what you save stays with you.

Yes — a full refund within 30 days of a payment, as long as the plan is unused (you haven’t saved anything that period). Once you’ve used the paid capacity, the current period isn’t refundable. See the Refund Policy.

The core — save, recall by meaning, and use across AIs — works today. On the Free plan, what you save stays with you.

No — each store is private to one account. Contextick is built around one person’s knowledge, and there’s no way to invite someone into your store.

Yes — a REST API for saving, reading and deleting straight from your own scripts (semantic search stays with MCP). Keys, authentication and endpoints are in the REST API docs.

When you only need short-term context inside one assistant and don’t care about the exact details later. If you switch AIs or need the specifics weeks on, that’s where Contextick helps.

Probably not. If you’ll never look something up again, there’s little reason to save it. Contextick pays off for knowledge you’ll want your AI to reuse.

If you like tending a system in Notion or Obsidian, keep it — that’s a real strength of those tools. Contextick is for people who’d rather not organize at all.

No. For images, video, or files you open directly, keep them where they are and save the link. Contextick is for the text you want your AI to recall.