A bit off the fintech trail but an interesting story that demonstrates how identifying a genuine business need, and being close to it, is all it takes to spark a startup.

Obviously, that is only the beginning and there are incredible challenges after that, such as funding, but the story I am going to tell shows clearly how tech startups should be built around a customer need.

I recently interviewed Jeff Kofman, who founded a tech startup known as Trint. Trint uses artificial intelligence to convert voice to text.

Kofman is a journalist and a successful one at that. He spent more than thirty years as a broadcast television journalist starting with the Canadian Broadcasting Corporation and then spent almost twenty years as a network correspondent in the US. In the US he worked at CBS Evening News and then New York Bureau and later spent 14 years at ABC News in Miami, covering Latin America. He also commuted to Baghdad for three years and then ended up in London for four years at ABC News London.

Kofman even won an Emmy covering the fall of Gadhafi during the Arab Spring and 2011. So he is  definitely a journalist and not what you would traditionally expect as a tech startup founder.

I have a particular interest in Jeff’s startup as it was created to solve a problem for journalists like myself. Although I do shorthand, for long interviews I prefer to record. This means re-listening to interviews and transcribing. This is laborious and time wasting. So after 30 odd years of doing this Jeff thought there must be a better way

He said he has lived the pain point like every journalist of having to listen to interview recordings and type out the quotes or the sound bites needed. “I met some developers that had done some work with transcription and I said ‘you know I would love to see if there is a way to use artificial intelligence and automate transcriptions.”

With his experience of the problem and the developers' skills combined they were able to create Trint. “I am solving a problem I know intimately and I understand what the solution needs to look like. I think that makes me  kind of a classic entrepreneur.”

Trint’s transcription technology is already being used by media organisations including The New York Times, the BBC, The Washington Post and Associated Press.

But it has moved way beyond the publishing industry and into higher education. He said, for example, sociologists, political scientists and anthropologists do interviews as part of their qualitative research. “They spend their research budget on transcription.” He said Trint has significant take-up by universities.

He  said marketing departments are also interested. "If you're doing marketing research for British Airways or Land Rover, you're doing qualitative projects. Why do you drive a Land Rover? Why do you think of a Jeep? What do you think of Virgin Atlantic? All of that is transcribed which is a very expensive and time-consuming process. People are telling us about user scenarios we didn't imagine."

He said Trint is also working with the legal industry and the police. “It is pretty much endless."

From the idea to the first product took about 20 months. “I left ABC News November 30th 2014 and I started working December 1st 2014 with a team of three developers. We launched the product in September 2016.”

“We had a prototype pretty quickly, in about three or four months, which we tested on journalists and they told us what they liked and what they didn’t like. We learned a lot about that”

He is surprised that he wound up doing what he is doing now. “If you had said to me 10 or 15 years ago that I would be running a startup and I’d be an entrepreneur and inventing software I would have said not a chance,” Kofman told me. “I’m a journalist through and through.”

By the way, I used Trint to do this interview. I was impressed with the accuracy and ease of use. You need to read it in a certain way and I suppose understand the subject matter, but all in all, it made writing this blog post much quicker and easier.

Automated transcription technology is improving rapidly.

Originally published here.