Cited answers
Every statement carries a numbered citation. Nothing in the answer stands without a passage from your documents behind it.
The assistant does not ask you to take its word for anything. Every answer shows where it came from, and the source is one click away.
Every statement carries a numbered citation. Nothing in the answer stands without a passage from your documents behind it.
Open a citation to read the passage it rests on, with its heading and page, then jump to that exact place in the document.
Each answer lists the documents it drew on and the pages it cited, so you can judge at a glance how well supported it is.
When your documents do not cover a question, the assistant says so. It does not fill the gap with an answer of its own.
Ask across everything you can access, or narrow a question to one workspace or to a handful of documents you pick.
The conversation keeps its context, so your next question can build on the answer you just read.
Also available: Ask from any document’s own page to question that one document alone.
For every question, LipiCore Documents first searches the workspaces you can access and picks the passages that best match, by words and by meaning.
Only those passages go to the AI model, with one instruction: answer from them alone, and cite each one you use. If nothing relevant is found, the model is never asked.
The model writes. Your documents decide.
The AI model you choose turns passages into a readable answer. What the answer says is fixed by what your documents say.
The assistant is the part you talk to. The same AI is already at work when a file is indexed, so documents are easier to find and easier to recognize before you open them.
On every document
Each document gets a short summary of what it covers, the keywords it is about, and a category from the list your organization defines. Nobody has to fill in a form.
Every passage is also indexed by its meaning. That is what lets a search for “getting paid back for a client trip” find the expense policy, and what lets the assistant pick the right passages for a question.
Consolidates guidance on joining, work schedules, leave and holidays, business travel, expense reimbursement and IT support. Specifies leave balances, the 2027 holiday calendar and expense filing deadlines.
Connect through MCP and your team can search your documents, read the passages that matter and get cited answers - in ChatGPT, Claude, Cursor or any other MCP-compatible tool.
One configuration in the tool. Nothing to install, nothing to export.
Explore LipiCore MCP Serverlist_workspacesSee which workspaces exist and what they holdsearch_documentsFind which documents are about a topicsearch_passagesFind where a document says somethingget_documentRead a document’s summary and detailsget_document_passagesRead a document passage by passageaskGet a ready, cited answer to a questionIt is the conversational interface inside LipiCore Documents. Ask a question about your documents in plain English and get an answer written from the passages that match it, with a numbered citation on every statement and a list of the documents and pages it drew on.
It is built not to. The assistant first retrieves the passages that match your question and the AI model is asked to answer from those passages alone, citing each one it uses. When no relevant passage is found, the assistant tells you so instead of answering. Every citation can be opened, so you can always check an answer against the document.
Yes. You can limit a question to one workspace or to the documents you pick, and every document page has an Ask button that opens the assistant on that document alone.
LipiCore does not provide its own LLM. You choose the AI provider and configure it in LipiCore, so the provider account, usage costs and model choice remain under your control. LipiCore Documents works with OpenAI, Amazon Bedrock, or a model you host yourself through Ollama. No LLM usage cost is bundled into LipiCore pricing.
Your files and the index stay inside your infrastructure. Two things are sent to the AI provider you configured, under your own key: the text of a document when it is indexed, to write its summary and index it by meaning, and, when you ask a question, the question with the passages retrieved for it. If you host the model yourself, nothing leaves your infrastructure at all.
Yes. LipiCore Documents includes an MCP server. Add its endpoint and access key to an MCP-compatible tool and that tool can search your documents, read passages and ask for cited answers. The built-in assistant stays available for everyone who does not work in an AI client.
A conversation lasts for your working session and belongs to the account that opened it. Conversations are not kept permanently, and one person never sees another person’s conversation.
Get in touch with our team to see how the AI Assistant in LipiCore Documents can turn your files into answers your team can check.