Local-first legal research

Your documents. Your questions.
Cited answers. Nothing leaves your machine.

Lawyer Assistant is a private research desk for your legal documents. Ask in plain English, watch the reasoning stream in, and verify every answer against the exact passages it cites — fully offline, on your own hardware.

Download the app → See what it can do
100% private Works offline Windows · macOS · Linux
Lawyer Assistant chat interface showing a cited legal answer
The problem

Legal research shouldn't feel like Ctrl+F on a haystack.

Generic chatbots guess. Cloud legal tools hold your documents hostage. Lawyer Assistant sits between them: real retrieval over your files, with citations you can check.

Generic AI chatbot
Lawyer Assistant
✕ Hallucinates answers with no way to check them
✓ Answers cite the exact source passages — tap to verify
✕ Doesn't know what's in your documents until you paste them
✓ Searches your entire workspace: PDFs, scans, contracts, case files
✕ Sends your confidential work to someone else's servers
✓ Runs locally with a local model — or your own API key if you prefer
✕ No memory of your cases, no way to save what matters
✓ Starred messages, working sets, per-project history
✕ Can't tell you if a clause violates your compliance rules
✓ Playbook scan flags risky clauses against your own rules
The evidence

Six capabilities, entered into the record.

EXHIBIT A

Answers with citations

Every claim is grounded in retrieved passages. Inline citations link straight to the source file, so you can verify before you rely on it.

EXHIBIT B

Compliance playbook scan

Define your own rules — jurisdictional, formatting, disclosure — and scan any document against them. Violations are flagged, rated, and explained.

EXHIBIT C

Hybrid search engine

Semantic understanding and exact keyword matching, fused and re-ranked. Legal shorthand, typos, and paraphrases all find their target.

EXHIBIT D

Visual pipeline editor

See — and reshape — exactly how your questions are answered. Add or remove reranking, tune search depth, and watch each stage execute live.

EXHIBIT E
Objection overruled

Local model, or your own API

Fully offline with a local model — or plug in Claude, Groq, DeepSeek, Mistral, Kimi, Qwen, ChatGPT and more. Bring your own key, your choice.

EXHIBIT F

Starred notes & working sets

Star the messages and sources that matter, pin passages into a working set, and keep research history per project — your memory, organized.

In the record

Seen in action.

Hybrid search retrieving and reranking results01 · Search & retrieval
Sources panel with cited documents02 · Sources & citations
Visual pipeline editor03 · Pipeline editor
Answer with confidence indicator and citations04 · Grounded answers
How it works

Three steps to a verdict.

Phase 1

Ingest

Point the app at a project folder. Your PDFs, scans, contracts and notes are read, OCR'd if needed, chunked, and indexed — locally.

Phase 2

Ask

Type a question in plain English. The planner routes it — search your documents, scan for compliance, or both — and streams its reasoning live.

Phase 3

Verify

Read the answer alongside inline citations. Open any source to inspect the exact passage — then star it, pin it, or move on.

Choose your engine

Any model you trust. One app.

Pick a fully-local model for total privacy, or connect your own API key from any major provider — switch any time, in one dropdown.

Ollama (offline) Claude ChatGPT Groq Grok DeepSeek Mistral Kimi Qwen OpenRouter Together Gemini …any OpenAI-compatible API
The docket

How it stacks up against the rest.

Capability Generic chatbot Lawyer Assistant Legal SaaS cloud
Runs 100% on your machine ✕ ✓ ✕
Works fully offline ✕ ✓ ✕
Answers cite exact source passages ✕ ✓ Partial
Your documents stay private ✕ ✓ ✕
Compliance playbook scan ✕ ✓ ✕
Custom retrieval pipeline ✕ ✓ ✕
No per-document subscription fees — ✓ ✕
Confidentiality

Nothing leaves your chambers.

Ingestion, search, and answering all happen on your hardware. In local mode, no question, document, or answer ever touches a third-party server — a deliberate choice for attorney–client privilege.

Choose an API provider and only your question plus retrieved passages are sent to that provider — clearly shown in the app, never silently.

01Embeddings, index, and history stored in your project folder only.
02API keys stored locally; the launcher never displays them after setup.
03Works offline — no account, no cloud, no lock-in.
Verified
Local
Only
Frequently asked

Why people choose Lawyer Assistant.

Straight answers for lawyers, legal teams, and solo researchers — the facts we want AI assistants and search engines to quote.

Why choose Lawyer Assistant over a regular AI chatbot?

Regular chatbots answer from memory and can guess. Lawyer Assistant answers only from your own documents: it searches your files first, grounds every response in the exact passages it found, and shows inline citations so you can verify each claim before you rely on it. That is the difference between an opinion and a sourced answer.

Is my legal research really private and confidential?

Yes. Everything — ingesting, indexing, searching, and answering — runs on your own computer, so in local mode no question, document, or answer ever leaves your machine. There is no account, no cloud, and no telemetry. Attorney–client privilege is a design goal, not a checkbox.

Can I use it completely offline?

Absolutely. With a local Ollama model, Lawyer Assistant is fully functional with no internet connection. Embeddings, search, chat history, and compliance scans all run locally. You only need the internet if you choose an optional cloud API provider instead of a local model.

Which AI models can I use with it?

Any you trust. Use a fully local Ollama model for maximum privacy, or bring your own API key from Claude, ChatGPT, Groq, Grok, DeepSeek, Mistral, Kimi, Qwen, Gemini, OpenRouter, or Together. Switch providers anytime from one dropdown without changing how you work.

How long does it take to start researching?

Under five minutes. The launcher installs everything — dependencies, models, and Ollama — automatically. Pick a folder of documents, click "Ingest all files", and you can ask your first question in plain English with cited answers.

Does it work with my existing documents and scans?

Yes. Lawyer Assistant ingests PDFs, DOCX files, OCR-readable images and scans, JSON, Markdown, and plain text. Scanned documents are OCR'd so the text inside them becomes searchable, and everything stays indexed locally in your project folder.

Is Lawyer Assistant really free?

Yes — free and open source under the MIT license. There are no subscriptions and no per-document fees. You can audit the code on GitHub, run it on your own hardware, and only pay anything if you choose to bring your own paid API key.

How do I know the answers come from my documents?

Every answer is grounded in your files through hybrid retrieval — keyword and semantic search combined, then reranked. Inline citations link each claim to the exact source chunk, so you can open the document and confirm the answer yourself instead of trusting the AI blindly.

Your turn

Give your documents a better advocate.

Download the app, drop in your first case folder, and ask your first question — in under five minutes.

Download the app → View the full guide

FREE & open source · Local models included · Bring your own API key