Switching Modes & Sessions
The AI Assistant has three modes — Everyday, Deep Research, and Agent. Exactly one mode is active at a time, and Everyday is the default. You can change mode whenever you like from the selector below the input bar.
What a mode applies to
The mode you choose applies to the next message you send — not retroactively to answers already on screen. Switching modes does not rewrite earlier responses in the conversation; it changes how the next question is handled.
A practical consequence: a single chat can contain replies produced in different modes. If you started in Everyday and switch to Agent, only the messages you send after that switch run as Agent investigations. The Everyday replies above stay as they are.
If you want the same question re-answered in a different mode, switch the selector and re-ask. The new mode handles the rephrased message; the original answer remains in the chat as history.
How mode persists
Mode persistence works on two levels:
- Per chat. Each existing chat reopens in the mode used for its last message. Open a chat that ended on an Agent memo and it reopens in Agent; a chat that ended on an Everyday reply reopens in Everyday.
- Globally, for new chats. Your last selection follows you forward. After you've switched to Agent or Deep Research, the next new chat opens in that mode — not Everyday.
Everyday is the system default for every new account, but once you've moved off it, the platform keeps your choice in place until you switch again. Glance at the selector before you send, especially at the start of a session or when reopening an older chat.
A single chat keeps the last mode it was used in. A new chat inherits the mode you most recently selected anywhere. Together: chats keep their identity, and your global selector follows you forward.
How the modes differ in output
Switching mode changes the shape of the answer:
| Mode | What you get |
|---|---|
| Everyday | A direct answer, no citations — copy-ready. |
| Deep Research | A memo-quality report with citations aggregated at the end — copy-ready. |
| Agent | A legal memo with inline citations linked to each source, plus a live investigation you can follow in the Thinking panel. |
Because the shapes differ, switching is also a way to change what comes back — not just how thorough it is. If a Deep Research report is more than you need, drop into Everyday for follow-ups in the same chat; quick clarifications stay quick and the report stays in the conversation as reference.
Combining modes in one session
It's normal to move between modes inside a single conversation. A few patterns:
- Triage in Everyday, then escalate to Agent. Use Everyday to orient yourself or get a quick read; switch to Agent for the part that needs to hold up.
- Deep Research for the survey, Agent for the case law. Lay out the legal landscape in Deep Research, then switch to Agent to validate the specific cases that drive the answer.
- Deep Research → Cite Check. Run Cite Check on every Deep Research report before you rely on its citations — Deep Research aggregates citations rather than verifying each against source.
- Agent for the memo, Everyday for the cover note. Once Agent has produced the cited memo, switch back to Everyday for the email or summary that goes with it.
Each message stands on its own under whichever mode is active when you send it. Earlier messages in the chat remain in their original mode — their citations, Thinking panel, and source panels don't disappear when you switch.
Verifying your work
Different modes ground citations differently. Agent verifies case-law citations against source as part of the investigation. Deep Research aggregates citations at the end, so run Cite Check on a Deep Research report to confirm every cited case is real, correctly formatted, and still good law before you rely on it. Everyday doesn't produce citations — if a question's answer needs them, that's a signal to switch modes.