General meeting
FastAI teammates gather quick opinions — the default meeting.
Not a single AI answering — an AI team deciding together.
Not a single AI answering — an AI team that reaches a conclusion together.
Roundroom is the AI meeting space where multiple AI teammates share different perspectives, review each other, and land on a conclusion you can act on right away.
Not an AI that talks — an AI team that meets.
How it works
How the AI team meets
Roundroom moves through three steps.
- 01
Share opinions
Each expert speaks in turn, from a different perspective.
- 02
Listen, rebut, refine
Each next expert hears the prior remarks and rebuts or builds on them, so the discussion flows.
- 03
Recommendation and decisions
A clear recommendation, what to decide in the real meeting, and next actions.
Three meeting modes
Secure meeting
SecureRuns in a dedicated enterprise environment on Google Cloud Vertex AI, keeping meeting data off third-party AI services.
Deep meeting
DeepDedicated AI agents research with web search and multi-step reasoning and remember past meetings, for a deeper discussion and conclusion.
16 expert agents
Team leads· 9
Planning Lead
Structures the agenda and clarifies priorities.
Engineering Lead
Checks technical feasibility and implementation cost.
Design Lead
Reviews user experience and visual flow.
Marketing Lead
Checks target, message, and channel fit.
Operations Lead
Defines execution steps and operational load.
Corporate Affairs Lead
Reviews organization, HR, and internal-policy impact.
Finance Lead
Reviews profitability, cost structure, and cash flow.
Legal Lead
Flags legal risk and the impact on terms and contracts.
QA Lead
Maps regression risk and verification scenarios.
Risk validation· 1
Risk Validation Lead
Finds failure modes and hidden risks early to limit losses.
Professional advisors· 5
Lawyer
Reviews contracts, disputes, and regulatory risk through a legal lens.
Certified Public Accountant
Checks financial statements, accounting treatment, and audit risk.
Tax Advisor
Reviews tax filings, optimization opportunities, and tax risk.
Patent Attorney
Reviews patent, trademark, design-right, and wider IP strategy.
Labor Consultant
Reviews employment contracts, labor relations, and labor-law risk.
Facilitation· 1
Meeting Moderator
Distills the discussion into a one-line conclusion, per-role summary, risks, and next actions.
Sample meeting
What a sample meeting looks like
Example: reviewing the pricing policy for a new subscription tier. Real meetings vary by agenda and team mix.
- Agenda
- Review the pricing policy for the new subscription tier
Each AI teammate's opinion
- Planning Lead
The customer problem is under-defined. We need to spell out who signs up and why first.
- Engineering Lead
As Planning said, we should narrow scope first — v1 is billing + access only, defer tier separation.
- Marketing Lead
Building on that, the core message is too abstract. We need a one-liner for ‘what changes for $X/month?’.
- Finance Lead
All good directions, but unit price, expected signups, and churn assumptions are missing — break-even first.
Round 2: pushback and final stances
- Planning: tune the user definition alongside the marketing message. — Final stance: Conditional
- Dev: v1 is doable if we assume a single tier. — Final stance: Conditional
- Finance: hard to proceed without P&L assumptions. — Final stance: Hold
Recommendation
Proceed with conditions — ship v1 with a single tier, then decide tier separation after four weeks of signup data.
Decide in the meeting
- Confirm v1 pricing and the single-tier setup
- Agree on the metric thresholds (signups / churn) for the tier-split decision
- Planning and marketing co-author a user-definition doc this week.
- Finance models a 12-month P&L with base and conservative scenarios.
- Engineering re-estimates v1 timeline and risk for billing + single tier.
- QA captures payment-failure and refund scenarios as pre-launch checks.
Roundroom helps surface issues and draft proposals before the real meeting. Humans make the final call.