Meeting Adjourned

A $4.3 billion AI market divided into three theories of execution.

AI meeting tools spent two years competing on transcription accuracy and summary quality, building a market that Grand View Research estimates at $4.3 billion in 2026. As recently as a year ago, the primary tension in the category was whether teams would tolerate a bot joining their calls. Three products that launched or expanded in recent months have moved the competition past that question. ZoomMate, Microsoft’s Copilot Cowork, and Granola all treat transcription as settled infrastructure, and they have each arrived at a different answer to the question that transcription left open: whose authority triggers the work that follows a meeting.

From transcript to ticket

Zoom launched ZoomMate on June 1, pricing the product at $20 per user per month and framing it as an “agentic AI work surface.” ZoomMate connects live meeting context to Salesforce, Jira, ServiceNow, Slack, and Workday, executing follow-up actions from what was discussed: creating tickets, updating CRM records, assigning tasks. Its “Complete” feature generates presentations, documents, and spreadsheets directly from conversational context, so that a meeting’s primary output is a deliverable.

Zoom’s CPO Russell Dicker described the product as connecting “what was decided to what needs to happen next” across every integrated system. The architecture reflects this belief. ZoomMate pulls context from Zoom Meetings, Phone, and Chat, and it also ingests conversations from Google Meet and Microsoft Teams, treating any meeting as a potential action trigger regardless of which platform hosted it. That cross-platform ingestion signals that Zoom values owning the conversational context above owning the meeting platform. The company has spent the past year reframing itself around what it calls a “system of action” vision, and ZoomMate emerged directly from that effort.

ZoomMate’s approach places it in a different segment from transcription-first tools like Otter.ai and Fireflies.ai, which capture and organize conversations but leave execution to the user. Those tools built the category and remain widely used, but the market’s center of gravity has shifted toward products that close the gap between discussion and action. ZoomMate, available in North America at launch with a global rollout planned, is Zoom’s bid to lead that shift.

Split along the seam

Microsoft’s Copilot Cowork, announced in March, approaches the same problem from the platform layer. Cowork draws on “Work IQ,” an intelligence layer that indexes Outlook, Teams, Excel, and the rest of the Microsoft 365 graph, treating meeting transcripts as one input alongside email threads, calendar patterns, and document activity. The product delegates tasks to background processes with defined checkpoints, running on mobile and desktop so that work progresses whether or not a laptop is open. The platform’s accumulated understanding of a user’s full work context determines what happens next, which means that a single meeting carries less weight in Cowork’s model than it does in ZoomMate’s.

Granola, which raised $125 million at a $1.5 billion valuation in March, locates authority with the individual user. The tool transcribes locally from the user’s machine without a visible bot joining the call, and it augments the user’s own notes with AI-generated detail and context. Granola launched MCP support in February, connecting meeting context to Claude, ChatGPT, and other AI tools, but the user decides which context flows where and what actions follow. Revenue grew 250% in the quarter before the March raise, driven largely by enterprise customers who adopted the product for its commitment to human-directed workflows.

The three positions map to different organizational assumptions. Sales-led and client-facing teams, where meeting-to-action latency carries a visible cost, form the natural constituency for ZoomMate’s approach. Organizations deep in the Microsoft 365 ecosystem will find that Cowork’s broader orchestration fits their existing workflow gravity. Granola’s enterprise client list, which includes Vanta, Gusto, Asana, Cursor, and Mistral AI, skews toward companies that hire for individual judgment and tend to resist tools that act autonomously. The alignment between tool philosophy and customer philosophy appears self-reinforcing.

Convergence and coexistence

The meeting AI market’s three-way split previews a dynamic that will likely recur across enterprise AI categories as agentic tools multiply. Whether a company centralizes or distributes decision-making authority shapes which tools feel natural to its teams. Features and pricing will converge as competitors copy one another’s integrations, and ZoomMate and Copilot already sit at comparable price points. Granola’s $1.5 billion valuation on a meeting-notes-plus-context pitch suggests that investors see room for all three approaches to coexist. The choice of meeting AI tool will, for most organizations, reveal more about how they already make decisions than about which product offered the better demo.