Unmeet is the product we're building to give them that data — and the permission to act on it. The plan: a pipeline of AI agents that read the real state of work across the tools your team already uses, score every meeting on the calendar, and learn from each one they see. The question we want it to answer: is this meeting earning its place today?
Unmeet is at the concept stage. The architecture below is the current plan, not the shipped product. We're working with design partners to figure out what should actually get built first.
Plenty of tools record what was said. We think the bigger question is whether the meeting should have happened at all. That's a different category, and a different value proposition.
Our plan: Unmeet's agents don't sit in the meeting. They sit in the tools your team already uses — project boards, code platforms, team chat, ticket systems, documents, calendars — and read what's actually getting done. Then they compare that against each meeting's purpose, score the gap, and recommend what to do about it.
Unmeet doesn't ask your team to adopt a new platform or install a new tool. It reads what's already happening across the systems your team uses and tells leadership which meetings are still earning their place. Four shifts we're aiming for from cycle one.
Most meetings are 80% already covered by what's getting done across the stack. Unmeet finds the 20% worth talking about and shrinks the rest. A 30-minute standup becomes 8 focused minutes. A weekly review gets cancelled outright.
Skipping a meeting isn't where Unmeet stops. Each attendee gets a specific thing that would move work forward — pulled from what's actually blocked on their plate. Guilt becomes momentum.
A single view shows how much synchronous time each team is spending — and how much of it the data trail says wasn't needed. The cost of meeting waste stops being invisible.
Unmeet reads from the tools your team already lives in — project boards, code, chat, calendars, docs. No new app, no new login, no new ritual to enforce. The team doesn't even need to know it's there.
Under the hood: connectors are built on MCP — the open Model Context Protocol every major AI vendor adopted in 2025-26. One standard, no vendor lock-in, read-only by default, SSO-authenticated. Your IT team will recognize it. Your end users will never need to.
One number on the leadership dashboard answers the only question that matters: is this meeting earning its place? Four signals feed it, scored together. The weighting will calibrate to your team over time.
If every sprint item has visible progress logged in the project board, the "what did you do yesterday" portion of the standup is already answered.
If a developer filed an access request four days ago and the approver hasn't responded, the system already knows the blocker — it doesn't need a human to say it out loud.
How much of what would be discussed is genuinely new versus already documented in tickets, comments, messages, design docs? A meeting where 90% is already visible scores very low.
Are there cross-functional decisions still unresolved that genuinely require synchronous discussion? Or are they decisions one person could just make and post?
The confidence score drives one of three actionable recommendations. Most meetings, we think, are mostly already covered by the data trail with a smaller slice worth talking about live. Unmeet's job is to find that slice and turn the meeting into focused time on what matters.
Everything the meeting was meant to cover is already addressed by the data trail. Unmeet recommends canceling — and redirects each attendee to higher-value work they should do instead with the reclaimed time.
Most of what the meeting was meant to cover is already addressed, but specific items genuinely need human discussion. Unmeet generates a focused agenda — the actual unresolved questions, who needs to answer each, and the context to skip — and cuts the meeting from 30 minutes to 8.
Significant unresolved decisions, cross-functional dependencies, or genuinely ambiguous situations that need real-time discussion. Unmeet confirms the meeting is justified — and may still pre-load a prioritized agenda to make it sharper.
Why three tiers matter: most meetings get shorter and sharper — that's where the value sits. Trimming a 30-minute meeting to 8 minutes, focused on the actual unresolved questions, is a much easier conversation to have with a team than skipping it entirely. The data shows where the value lives, and the team makes the call.
A productivity multiplier tells you what to do with the time instead. Unmeet's agents already know what's blocked, what's stuck, and what would actually move work forward. When the system recommends skipping a meeting, the idea is to show each attendee exactly what would be more valuable to spend that hour on.
This is the difference between "you wasted six hours this week" and "you saved six hours this week — and here's what your team did with them." One emotion is guilt. The other is momentum.
The recommendations should get more accurate for your team with every meeting Unmeet sees. The model of how your teams actually meet, decide, and execute gets built one cycle at a time.
Score, recommendation, and focused agenda (if shortening) are delivered to leadership and attendees before the meeting starts.
If a transcript is available from your conferencing tool or an existing transcription assistant, Unmeet ingests it. No new tool in the room.
Was the prediction right? Were the “redundant” items actually redundant? The model updates from real outcomes — yours, not someone else's.
The longer Unmeet runs in your environment, the more accurate its recommendations get for your specific teams. The calibration history — the model of how your teams actually meet, decide, and execute — gets built one meeting at a time, and it stays yours.
You decide how much autonomy to grant. Start conservative and expand as trust builds, or stay conservative forever. Every action is auditable in every mode.
Unmeet surfaces recommendations to leadership and meeting owners. Nothing happens automatically. A human reviews each skip / shorten / keep call and decides whether to act on it. The recommended way to start — lowest risk, highest trust-building.
Shorten and Keep recommendations apply automatically — focused agenda sent to attendees, durations adjusted in the calendar. Skip recommendations still need explicit human approval before a meeting is canceled. The middle ground most teams settle into.
All recommendations apply automatically, with a human override always available. Requires a calibration period (at least 20 feedback cycles) and an explicit organizational opt-in. For teams that trust the model and want the lift.
Audit trail is always on. Every recommendation, every human decision, every applied action, and every outcome is captured — for compliance and retrospectives.
Two examples of the meetings Unmeet is designed to evaluate. Scores below are illustrative; the live product will calibrate to each team's actual patterns.
Unmeet is built for leaders who can see the cost of meeting waste on their P&L. Engineering, operations, strategy — anywhere the labor cost of unnecessary synchronous time is a real number.
Owns velocity and engineering cost. Wants to know which ceremonies are taxes versus genuinely useful.
Looks across the whole org. Sees the hours lost to weekly reviews, syncs, and cross-functional meetings — and wants them back.
Responsible for org effectiveness. Wants the data to support changes they already suspect are needed.
Engineering managers, product directors, ops leaders — anyone whose team can win back hours for higher-value work.
We believe in showing our work. Below is the framework, the inputs, and the math, so you can plug in your own numbers and decide for yourself what the reclaim is worth to your organization.
The pattern we expect to see is consistent: even a modest meeting reduction returns more capacity than the tool costs. The full upside — what your team accomplishes with the time back — sits beyond the math entirely.
Unmeet is a concept right now: a thesis, a designed system, a clear plan. Verdict is the product we built first, and we won't split engineering attention until Verdict is established with paying customers.
We're seeking design partners now — organizations willing to help shape what Unmeet becomes. If meeting waste is on your radar, the waitlist is the first step.
If your team's calendar is heavier than it needs to be, we'd love to talk. Even at concept stage, we'd value your input on how Unmeet should be shaped for your environment.