How one CEO turned meeting waffle into a competitive advantage

My interview with Sam Liang, co-founder and CEO of Silicon Valley tech firm Otter.ai, is like nothing I've done before — but maybe it's the future.

We meet at a café inside the lobby of a London hotel. He opens the Otter app on his phone, lays it on the table and starts recording. He opens his laptop, where he's also logged in to Otter.

Five minutes into our conversation, he is already interrogating what we've discussed so far. In near real-time, the transcript (which has travelled across the Atlantic and back and seems unfazed by the hotel’s muzak) appears on the laptop’s browser. He uses a sidebar to ask his questions, receiving summaries of our key points.

He's been discussing his company’s big idea, the Conversational Knowledge Engine, which he thinks is the future of the workplace. But it comes with an important caveat that not everyone will agree with: all work meetings should be recorded.

Otter.ai's workforce is its own Customer Zero, so it’s been following this strategy by using this frontier tech for many years. “I expect [staff] to record everything all the time and I can show you on my laptop,” he tells me.

This includes both internal and client-facing calls. The idea is that all data can be analysed by AI intelligently and logically, without human bias or fallibility. He shows me how he can observe a customer sales call happening in real time.

“Otter actually knows everything about the business,” Liang tells me. “[The idea is to] make knowledge as broadly shared as possible, because a lot of businesses are slow and inefficient because of information silos. Companies spend more than 50 per cent of the payroll dollars to pay people to go to meetings, yet most of that conversational knowledge is not even captured. So you can imagine how much waste this creates.”

Teams not talking to each other and ‘information silos’ – that’s certainly two things I hear a lot from businesses. And I’ve experienced it firsthand myself too, of course, as an employee at various places.

Liang grew up in China. He is a computer scientist with a PhD from Stanford and an ex-Google employee who left to start his own tech ventures in Silicon Valley. He founded Otter.ai in 2016. I first encountered it as a journalist around that time as an early transcription tool, which was still novel to me back then - and very useful when it came to writing up interviews.

I ask Liang if he had to pivot when transcription became baked into so many AI offers, like ChatGPT or Google’s NotebookLM.

“Transcription was never the end goal,” he says. “Transcription was the very first step to convert voice to the written word. But the goal from the very beginning was to capture knowledge and organise and analyse it using AI."

His software service now is more like a company communications platform. It has an interface with channels and messages that resembles Slack or Microsoft Teams. Managers can use this platform to look at transcriptions of sales calls in real time, or check in on meetings, while quickly generating summaries on the fly. They can also feed thoughts into these events as they happen.

The idea is to have more joined-up thinking in the workplace. If customers are saying on calls there are certain product features they would like, for example, AI has the 360-degree overview that can act on that insight. It might prompt this information to be shared with the teams who need to know, like those in product development.

But what about the human side to this? As a journalist who is used to recording conversations for interviews, I'm always acutely aware of when the recording button is off or on. I'm also highly conscious of the quality of the recording and how much storage I have on my device (these days, my iPhone 15). This can be anxiety-inducing...

But I can tell that Liang is very unself-conscious about being recorded; it's as natural to him as breathing. The amount of storage needed to record text transcription is actually quite small, he points out, so he's not worried about the long ream of words we are generating being sent off to the cloud.

In the near future, writing will become a thing of the past, he believes, as obsolete an idea as turning your hands on a car steering wheel. (Indeed, he tells me that, as a long-term Tesla driver in California used to autopilot mode, he fears he has forgotten how to drive).

If he’s right, we will become more of an oral culture again, but with AI analysing everything and acting as our collective super-brain. The future is transcribed, as he sees it. Offices will be full of people whispering into their headphones, rather than furiously tapping away on keyboards.

Journalists like me are in the 1 per cent in society, so maybe don't get it, Liang points out, perhaps because I look sceptical. We are comfortable writing and expressing ourselves in words — most people aren't like that, he argues. Talking comes more easily to most people. And if AI can help you compose your thoughts, all the better.

But will office workers be comfortable having every word recorded in the name of AI-efficiency? Maybe people in the tech bubble of Silicon Valley have a blind spot there, I wonder? Though I suppose AI notetakers loitering in waiting rooms online are pretty common already and haven’t sparked a human revolt yet.

The AI revolution threatened to eat up software companies, as the ability to code the apps and services you want became much easier. Otter has elevated its transcription software to match the new possibilities AI ushers in.

It also has its own proprietary transcription tools, which is its protective moat. Its transcription software is fast and accurate and learns to adapt to particular speakers over time. But rivals have spotted the opportunity in this transcription space, including the likes of Granola, Fireflies and Wispr Flow.

Otter.ai’s challenge is to compete with rivals, maintain a system that can provide value to a company and integrate with other systems, like Salesforce or HubSpot, for example. Liang's company currently has annual recurring revenues of more than $100m from its clients.

As well as offering data to steer the company more efficiently and rationally, transcription also opens up the possibility of better personal feedback. This applies to the CEO too. Liang himself is of course 'on-record' all the time.

He uses AI for his personal development, like a coach. Sometimes he will analyse his own performance after a meeting. Or ask AI to give him feedback on his last 40 meetings. Did he talk too little? Did he listen enough?

Liang reflects: “When human beings give me feedback, sometimes I get emotionally embarrassed or sometimes I get too defensive. But when AI gives me feedback, I have nobody to be angry at. I can take it more objectively.”

Liang was in London to run the London Marathon. He tells me later he was privileged to witness a human break the 2-hour record and that it has inspired him in business. And as he points out with a smile, there is one job that AI can't touch: the athlete!