Deep Research
QuaeroLex thinks about the question, searches the matter from several angles, reads relevant sections from your documents and reasons before it answers.
Private legal document intelligence
Ask your case files — briefs, pleadings, witness statements, exhibits — in plain English. QuaeroLex finds the answer and cites it to the exact page, so you can verify it in seconds. It's built to take the legwork off your team — not to replace their judgement — and runs entirely on your own infrastructure, so privileged material never leaves your network.
Capability
Put a question to thousands of pages and get a cited answer in minutes — then open the source to verify it. Retrieval, reasoning and verification built specifically for legal documents, not a chatbot bolted onto a search box.
QuaeroLex thinks about the question, searches the matter from several angles, reads relevant sections from your documents and reasons before it answers.
QuaeroLex drafts a structured report from your own templates and styles — a first draft to edit and make your own, not a finished document to file. A starting point that saves you the blank page.
QuaeroLex is built to save you time, not to replace your judgment or hand you a finished report. It does the legwork and drafts a cited starting point — every passage linked to its source — so the work left to you is checking and refining, not searching from scratch.
How it works
Three steps, no specialist setup — and your documents never leave your infrastructure to do any of it.
Upload the matter's PDFs. QuaeroLex reads your documents and figures, converting them into structured text it understands.
Ask questions in plain English. The model answers only from the documents you provided and cites the sources for each answer.
Every answer arrives with citations you can open and check on the source page — so the final call always stays with you. The system flags anything it can't substantiate for a closer look.
Security & confidentiality
QuaeroLex is designed for documents that can't leave the building. It runs inside your own perimeter, and is architected so that no case data — and no metadata about it — is ever sent anywhere you don't control.
Deploy on-premise or in your own private cloud. You own the box; we never need access to it, and there is no managed service in the loop.
Answers are generated by a local model on your own machines. Nothing is sent to an external API — not for inference, not for embeddings, ever.
The application doesn't phone home. No usage analytics, no crash reporting, no background calls — it works fully air-gapped if you need it to.
Per-user permissions scope who can open which matters, and a complete record captures every question asked and every source it relied on.
Documents and their indexes are encrypted on disk, and all traffic within your deployment is served over TLS between every component.
Nothing you upload is used to train any model. Close a matter and its documents and index are removed — the data lifecycle is entirely in your hands.
Need to satisfy your IT or risk team first? Request a demo and we'll walk through the deployment and security model in detail.
First-draft reports
Point QuaeroLex at a matter and it drafts a starting point — allegations, credibility, the defendant's case, liability and quantum — each paragraph grounded in the file and cited as it goes. It's a head start to edit, check and make your own, not a finished advice to send.
Questions
The practical questions that come up when privileged material is involved. If yours isn't here, ask us directly.
No. QuaeroLex is a tool for qualified lawyers and barristers, not a replacement for them. It does the reading and retrieval and produces cited first drafts to save you time — it doesn't deliver finished, send-ready work. A competent person always reviews, edits and takes responsibility for the final product.
It answers only from the documents you provide, and every claim is cited to a source page you can open and check. Anything it can't substantiate is flagged rather than guessed at. It's built to speed up your own review, not to replace it — you confirm each point against the source, and the final judgement stays with you.
A capable large language model that runs locally on hardware you control. Both inference and document embeddings happen on your own machines — nothing is sent to OpenAI, Anthropic, or any other third-party API.
It runs on a server you control, on-premise or in your own private cloud. We'll recommend a specification sized to your matter volumes as part of the walkthrough, and help you stand it up.
No. Documents, indexes and answers all stay inside your perimeter. There's no telemetry and no phone-home — it can run fully air-gapped if your environment requires it.
Per-user permissions govern who can open which matters, and a complete audit trail records every question asked and every source behind each answer — so use of privileged material is always accountable.
Because everything runs on your own infrastructure, your data stays in your jurisdiction under your existing controls and retention policies. Nothing is processed by an outside service, and nothing you upload is ever used to train a model.
Tell us a little about your practice and we'll arrange a walkthrough — including how a self-hosted deployment would work inside your firm.