Transcription bots move beyond note-taking into running meetings
AI transcription tools have become so embedded in video meetings that workers panic when they go offline. When Granola, a popular transcription service, experienced an outage in October 2025 due to an AWS failure, users flooded social media expressing distress over losing access to their "second brain."
The market is responding to this reliance with aggressive growth projections. The AI transcription industry is expected to expand from $4.5 billion in 2025 to $19.2 billion by 2034, with over 130 companies now competing in the space.
From passive listeners to active agents
Fireflies, founded in 2016, serves 800,000 users across tens of thousands of teams. The company initially focused on improving transcription accuracy but has shifted toward becoming a knowledge management platform that distributes meeting information across workflows.
"It's now knowledge orchestration, where we're taking all the knowledge that happens in conversations, and we're putting it in the places where you work," said Krish Ramineni, Fireflies' founder and CEO. The company operates 90-plus integrations with tools like Notion, Slack, and Salesforce.
Notetaker companies are now investing heavily in AI agents that move beyond passive listening. "The notetakers that are winning these days are doing more than just being those passive listeners. They're actually being action doers," said Natalie Rutgers, VP of product at Deepgram, which powers tools like Granola and Otter.
Read.ai, founded in 2021, has built specialized agents for healthcare, construction, and sales. In sales, the tool extracts details from customer calls and automatically pushes data into CRM systems like Salesforce or HubSpot-eliminating manual data entry.
The compute cost problem
Running transcription at scale is expensive. Processing hours of meetings daily for every user requires significant compute resources, and speech-to-text inference costs roughly five to six times more than text-only models.
"Demand for compute is going to be huge as these services expand across organisations, leading to more and more inference calls," said Michael Stothard, principal at Firstminute Capital, a venture investor in Granola.
Ramineni estimates that roughly a billion knowledge workers globally average a few meetings per week. "At some point every meeting on the planet will be captured by an AI system like Read. There will be a lot of increased demand for these foundational capabilities," he said.
Fireflies has addressed cost pressures by optimizing its model stack and using five different model providers to achieve efficiency. The company has been profitable since 2023.
What comes next
Speech-to-text technology is expanding beyond meetings. Jack in the Box now uses Deepgram's models to take drive-through orders, and humanitarian organizations deploy voicebots to triage emergency calls during disasters.
For notetakers themselves, competing visions are emerging. Some companies dismiss flashy features like video avatars as premature. Others are pursuing something more radical: a product designed to fully replace humans attending meetings.
Your membership also unlocks:
