At many meetings these days, it can feel like there are more AI note-taking and transcription bots than humans. There seem to be dozens of options these days, but one I've started seeing more and more of is Fathom. The company was one of the early players, launching in 2020. Fathom then raised a $4.7 million seed round in early 2022 and today announced a $17 million Series A round led by partners at Telescope. Notably, $2 million of the $17 million was provided by Fathom users via crowdfunding on Wefunder.
The company claims that over the past two years, it has seen a 90x increase in revenue and a 20x increase in usage. This is likely from a relatively low baseline, and while the company doesn't disclose exact active user numbers, Fathom, which offers a generous free plan, says that more than 8,500 companies are currently using its integration with HubSpot.
Image credit: Fathom
“We've always built Fathom with the expectation that AI would be really great,” the company's CEO and co-founder, Richard White, told me. “When we started in 2020, AI didn't exist yet. But we were going to really focus on the things that seemed hard — getting distribution, building a really reliable infrastructure, and an easy-to-use product — with the expectation that when AI got there, we'd be able to embed AI into this easy-to-use product that was already widely adopted, and it would evolve from good to great.”
One thing that has always set Fathom apart is the company's reliance on proprietary models, or at least its own tweaked versions of open models. Fathom has its own team that works and experiments with models, and as White pointed out, working with models is very different from a typical engineering project.
“Their deliverables aren't features, they're actually specs,” he said of this team, “and they have a failure rate. It's not like engineering, where you put something on the roadmap and it's done. Right now it's 50% of the time, and if it's not good enough, let's check back in six months. I think we had to rethink our product development process a little bit.”
Over time, the team added a number of new features, including action items and automated drafting of follow-up emails, a chatbot called “Ask Fathom,” team-oriented features, etc. Most of these advanced features are only available on paid plans, with monthly plans starting at $19.
Image credit: Fathom
Many of the larger conferencing services have begun to offer their own take on Fathom's efforts, but White doesn't seem too worried about that. “Our vision is to have all meetings in one place,” he said. That broader vision includes being the information hub for a company's executive team, something White said is hard to achieve when supporting only a single conferencing platform.
Part of this vision, and part of what the new funding round will help Fathom's efforts, is to help users not only with meeting notes and action items, but also with many of the post-meeting chores. It's started down this path with its automated action item and follow-up email features, but the idea here is to build out more integrations and use AI agents to perform more of these tasks, interfacing directly with CRM systems, for example.
White also noted that there's a ton of data currently untapped in meetings that could provide executives with more environmental information over time. After all, no one can attend every meeting, but White envisions a more proactive system that could alert decision makers when, for example, the sales team faces a question for which they can't come up with a good answer, or when a competitor's name is suddenly being mentioned more frequently in meetings.
I tried Fathom out on a few meetings this week and was generally impressed with the quality of its transcriptions, but especially the meeting summaries. Whereas other tools (such as Otter) seem to try to shoehorn every meeting into a somewhat rigid flow that's primarily tailored for sales users, Fathom simply provides a handy summary that smartly creates chapters for every meeting.
Below is a summary of our interview with Fathom: The only mistake here was not setting up a follow-up meeting.
Image credit: Fathom/TechCrunch
Source link