Deep Research
Some questions don't have a one-line answer — a regulatory analysis, a statute read across jurisdictions, a survey of secondary sources. Deep Research is the mode for those. It takes one comprehensive pass: it pulls from online legal resources and your matter, synthesizes across them, and returns a structured, memo-quality report. It is slower than Everyday on purpose, because it is doing the work a quick answer skips.
What you get
Deep Research returns a structured report, not a chat reply:
- A direct answer to the question.
- Supporting analysis organized by sub-topic or element.
- Aggregated citations to the sources it relied on, gathered at the end.
- A note on open questions or areas where the law is unsettled.
The citations are gathered at the end rather than threaded inline, which keeps the report copy-ready — you can paste it straight into your own work without stripping anything out.
How it works
Ask a research question
Select Deep Research in the mode selector and ask, for example: "What are the defenses to a breach of fiduciary duty claim in New York?"
One comprehensive pass
Deep Research draws on a few different sources — the documents in your matter, public legal resources and government sites on the web, and online federal and state statutes — and synthesizes across all of them in a single pass. There is no back-and-forth investigation loop; that step-by-step pursuit of leads is what Agent mode does.
A structured report
The result is the memo-quality write-up above — answer, analysis, aggregated citations, and open questions.
Verifying the report
Because Deep Research gathers its citations at the end rather than verifying each one against source as it writes, run Cite Check on the report to confirm every cited case exists, is correctly formatted, and is still good law before you rely on it. Treat the two as a paired workflow: Deep Research produces the report; Cite Check signs off on the citations.
For a question that turns heavily on case law, Agent is the better starting point — it verifies case-law citations against source as part of the investigation and threads them inline.
When to use Deep Research
| Scenario | Everyday | Deep Research |
|---|---|---|
| Quick case lookup | Recommended | Overkill |
| Single-issue legal question | Good | Better |
| Multi-element analysis (e.g., "elements of fraud in Texas") | Limited | Recommended |
| Memo or brief preparation | Insufficient | Recommended |
| Cross-jurisdictional comparison | Limited | Recommended |
When not to use Deep Research
- You need a quick answer or a short draft — Everyday is faster and the right shape.
- The question turns sharply on case law and the answer has to hold up against a court — Agent's inline-cited memo and live verification fit the job better.
- You're working a matter with hundreds of documents and need a document-by-document read — Agent reads selectively across large matters; Deep Research is built for a single synthesis pass, not a deep document trawl.
Using Deep Research with the other modes
- Deep Research → Cite Check. Always run Cite Check on a Deep Research report before relying on its citations. The two were designed to work together.
- Deep Research → Agent for the case law. Use Deep Research to map the legal landscape on a topic (statutes, regulations, secondary sources), then switch to Agent to pull, verify, and analyze the specific cases that drive the answer.
- Everyday for follow-ups. Once a Deep Research report is in the chat, drop into Everyday for quick follow-up questions — clarifications stay fast, and the report stays in the conversation as reference.
Switching modes mid-conversation does not rewrite earlier answers — your Deep Research report stays exactly as written when you move to Everyday or Agent for the next message. See Switching Modes & Sessions.
Working with the report
Download any report as PDF or Word (DOCX), use Save this output to Notes in My Desk to keep it with the matter, or copy it straight out.