Overview

The War Room is a multi-agent council inside the AI Console. You seat a table of two to fifty AI panelists, set a topic, and watch them debate it in structured rounds while a host keeps order and closes with a written synthesis and a headline verdict. This page explains what the War Room is, where to find it, who can use it, and the key ideas you will meet across the rest of this section. It is available on the Pro plan and above, and it is rolling out to eligible accounts in waves.

What it shows

The War Room opens from the AI Console, where the entry point is labeled War Room. The screen has a left configuration rail with three numbered steps, Topic, The Table, and Format, and a right stage that previews the round table before you start and becomes the live theater once the council convenes. A slim header carries New session, Presets, and History controls, plus a live readout of the credits spent and your remaining balance.

You assemble a lineup, choose how the panelists should argue, and press Start session. Each panelist takes its turn, can pull live research while it speaks, and the host writes the conclusion when the rounds end. Every session is saved, so you can replay it, ask the same panel a follow-up, or share it.

Key concepts

The rest of this section documents each idea in depth. The following terms are the ones to know first.

Session: one run of the council, from the moment you press start to the closing verdict. Each session is saved for you to reopen or continue. See The live room and the verdict.
Council: the group of panelists you assemble for a session. See The table and its seats.
Seats: each panelist sits in a seat at the table. A seat is one of your saved subagents or a one-time agent you create for a single room, and a room seats two to fifty. See The table and its seats.
Host: the orchestrator that leads the room. It opens each round and writes the closing synthesis. GNG Analyst leads by default, or you can put one of your own subagents in charge. See Formats, modes, and settings.
Rounds: how many times each seat speaks. Panelists take turns in seat order or all at once, for a set number of rounds or until the host judges it has heard enough. See Formats, modes, and settings.
Modes: the tone and structure of the debate, from an adversarial cross-examination to a collaborative analysis or a split between bullish and bearish seats. See Formats, modes, and settings.
Verdict and synthesis: when the council finishes, the host writes a full synthesis and lifts its lead sentence to the top as the verdict, so the takeaway stands out. See The live room and the verdict.
Deep Research Report: the written research note every finished council produces, opened from the finished room's Results view. See The Deep Research Report.
Spend cap: a hard ceiling on a room's spend in GNG credits. When the running cost reaches it, the room stops and delivers a closing summary, so a session never runs away from you. See Spend cap and credits.

Use cases

Stress-test a thesis. Seat an adversarial panel and let it cross-examine your reasoning from several angles, so the weakest points surface before you act.
Argue bull versus bear on a position. Split the table for and against one name, then read a balanced verdict that weighs both sides.
Convene a mock analyst team on a sector. Seat specialists that each own one angle, for example fundamentals, technicals, and sentiment, and let them collaborate into one combined read.
Keep a debate on record. Every session is saved, so you can reopen it, ask the same council a sharper follow-up, or share a read-only link with a colleague.

GNG Research provides equity research and educational tools, not investment advice. Nothing on the platform is a recommendation to buy or sell any security. Do your own research and consider your circumstances before making any investment decision.