Jeff Kofman is an Emmy-winning foreign and war correspondent who spent three decades working for major broadcasting networks such as ABC and CBC, reporting from across the globe covering some of the biggest news events of recent years, including the Iraq war, the Arab Spring, and the Libyan Revolution. In 2014, he swapped war zones for Zone 1, and founded Trint, the AI automatic transcription service that is saving his fellow journalists from hours of tedious, time-consuming work. Figaro Digital sat down with Kofman at Trint’s trendy East London office to discuss his motivations, experiences, and startup success.
FD: What prompted you to end your well-established career as a broadcaster and dive into a startup in London?
JK: I had been a broadcast journalist for more than 30 years. When I began in my early 20s, I’m not sure I had a conscious bucket list, but I did so much more than I could have ever reasonably dreamed. From Canada to the US to Latin America, I covered everything from penguins in Patagonia and blue whales in southern Chile, to the drug wars, the horrors of Haiti’s earthquake, and the Iraq war. After 30 years I felt that I had the energy to do something different, that I was getting too comfortable where I was and needed a change. I did not plan this. This completely happened by surprise.
I was London correspondent for ABC News in my last position and I decided I didn’t see myself coasting to retirement in that job. A friend took me to a conference for media developers at the O2 called MozFest. He introduced me to a team that was doing something on a prototype basis around manual transcription, tethering the audio to the text and making it searchable. I was kind of blown away and asked: “couldn’t you put automated speech in there, and find a way to fix it if it makes an error?” And that’s how Trint was born. It was just a chance meeting. I remember saying to them: “this is the future.”
I fit a classic profile of an entrepreneur – I’m solving a problem that I know intimately. And that’s a fairly classic startup role; it doesn’t tend to be people who are in their 20s, it tends to be people who have some life experience. What’s interesting is that the product I imagined that day is pretty close to what we released. That’s because I spent thousands of hours of my life transcribing and I intimately understood what the solution had to look like for it to work for me. The world needed to find a way to make voice easily and affordably searchable. Everything in my career around the technology of journalism had changed since I began in the 1980s; transcription also needed to be brought into the 21st century.
FD: How has your background as a journalist prepared you for a life in business?
JK: In some ways it’s prepared me a lot, in other ways it hasn’t prepared me at all. There are a couple of things that I just learned intuitively over the years. I’m totally comfortable asking stupid questions. I’m resourceful because I know how to ask focused questions, and I’m also very comfortable saying “I have no idea what this is, please tell me.” A lot of people are embarrassed by what they don’t know – I can laugh at it. I’m generally very decisive because when you work in difficult situations under immovable deadlines, it forces you to be really decisive, and to accept that you will make mistakes. But that’s OK. You make an informed decision and you accept that sometimes it will be wrong. That’s way more empowering than being indecisive and making no decision. You have to be decisive that way and in a lot of ways my journalism helped me prepare for that.
Where it didn’t was in the financial side, and the business skills have been really challenging. When I first tried to work on Microsoft Excel, I had no idea what I was doing and all I got was ‘#####’. I joke that I wanted to curl up under my dining table – send me back to a war zone, please! I’ve managed to build a really strong board that’s helped me a lot, and now that we have a management team and a finance manager I don’t feel as overwhelmed. But I did a lot of it on my own for the first two years. It’s quite amazing when I think about it – how did I manage to not completely mess it up?
One of the things about being a war correspondent is that you end up in some pretty sticky situations, some of which are quite terrifying, so you need to know how to stay calm the worse things get. That’s quite a useful thing when you run a startup as you’re constantly running into landmines. The advantage now is that the landmines are metaphorical. In my previous job they weren’t.
FD: What digital innovations benefited your previous career?
JK: I began in television in Canada in the 1980s on manual typewriters, so when the first word processors came out I thought that was a gift from the gods. I was never a great typist, I’m fast but not perfect. Writing for television you have to have a clean script to read on the teleprompter, and when I wrote for newspapers in my early years, they were scanned optically so it had to be perfect. A lot of paper was lost as I retyped and retyped. So when the first version of DOS Text Editing came online, I was tearful in glee! That’s all I wanted – I didn’t really care whether it could do anything more than that. I also had the very first Mac and it was the most liberating thing.
FD: What have you found hardest about setting up your own company?
JK: It’s really hard. It takes perseverance, drive, tenacity – all those clichés are absolutely true. There are often a thousand and one reasons to give up. You have to believe in what you’re doing and you have to care about it. If you’re driven by money – good luck to you. I can’t imagine being solely driven on this mission by money. You have to believe that what you’re doing has the potential to be transformative. It excites you, and through that you can excite your team.
I think it’s also the relentlessness of it. When you’re a CEO there are always issues. Managing 30 people, there isn’t an employee in the office who I don’t think is terrific, but that doesn’t mean it’s not challenging. People have personal issues, they get sick, life happens – and you just have to deal with that. Also, the sheer volume of decisions can be exhausting. I belong to a cycling club and I go cycling on Sunday mornings. I never lead one of the rides. I spend my entire week leading. The last thing I want to do on Sunday is tell people where to turn and when we’re going to stop. I love to follow on Sunday mornings.
JK: Traditionally, the only way to get the content out of an interview, speech, lecture, keynote, news conference, or meeting, is for somebody to put on headphones, (either you, a colleague, or an expensive third party), and hit play, hit stop, type it out, rewind, hit play, etc. It generally takes three to four times the length of the recording to transcribe it accurately. When I began in the 1980s on manual typewriters with a cassette recorder in my hand, when videotape was just being introduced, that’s how I did it. Thirty years later, it’s still how it’s done, except we use digital recorders. I asked the question: can’t we use artificial intelligence to do the heavy lifting? Essentially you take a recording, either audio or video, you upload it to Trint, and minutes later we spit out a machine generated transcript. It’s a first draft of the transcript, but if you give us clear audio it will be 95 per cent or more accurate. Because I’m a broadcaster, my voice comes out really accurately; Theresa May, Barack Obama, Donald Trump come out really well too. If you start to get heavy accents or weaker audio, it will degrade.
What we give you back is not just the text, because you can get that from others. The problem is that text is by definition flawed, which means that for me as a journalist, for you as a marketing person, for a doctor, for a government worker, for a lawyer, an inaccurate transcript is useless if you can’t verify it. A text document that is separate from the audio is incredibly difficult to verify. We married a text editor to an audio/video player, essentially creating a word processor for audio and video, making it instantly searchable. You can then listen to it, verify it, and if necessary correct it. So if I’m looking for a specific reference in an interview, I can search a keyword, and I can find that there it is, hypothetically thirty three and a half minutes in, and I can hit play. If there’s a name that got messed up, or if somebody stumbled on their word, I can clean it up. And in literally 20 seconds I’ve got that quote. Anybody who knows what transcription involves knows that in an hour-long interview that would be at least 20 minutes to find, if not an hour or two depending on the circumstance.
FD: Trint has been going for just over three years, and you’ve been in the market for a year and a half. What are your goals for the future?
JK: We want to create a new language for voice. We live in a world where 80 per cent of the communications online are in audio and video form. It’s not searchable. We are introducing a whole new way of accessing recorded content and we have really big ambitions, so that’s what we’re starting to do. It’s about speeding up the discoverability and shareability of content. We began as a transcription platform, we’ve now moved into a production platform, and ultimately we will give people the options for publishing as well.
FD: Despite being a seeming niche product, automatic transcription is actually a highly competitive market. What have you learnt from your competitor’s flaws or mistakes?
JK: It’s all about workflow. It’s fine to have cheaper prices, it’s fine to have a product that claims to do this and that, but if it doesn’t have good accuracy and it’s not easy to use it doesn’t matter. Through really rigorous user testing at a very early state, we figured out how to make Trint so intuitive that you don’t need to use the help pages. Ease of use is critical. We’ve all had experiences on the internet, when you go to use something, you get stuck, you try again, give up, and never go back. It’s bad UX and bad UI (user interface). Our core UX and UI are so intuitive people love it, because it doesn’t require anything more than just basic logic.
FD: What do you say to people who are apprehensive that AI might take their jobs?
JK: I suppose in some circumstances it may take away jobs but we don’t see it as doing that. We see it as liberating you to do your job. A journalist is not a stenographer. A good journalist listens to their interviews and reviews them. But that doesn’t mean to do your job well you need to type it out. If you can listen to it in real-time, or search it and listen to the salient moments, you are liberated to spend more time creating content and less time doing a menial, mechanical job. We’re not causing the layoffs and cuts in newsrooms, what we’re doing is saying with fewer people you can actually be more productive. In this case I don’t think AI is the villain – AI is the liberator. The important thing to understand with Trint is that we don’t claim the machine generated transcript to be the end product. There is human intervention required – it’s just a heck of a lot less than it was with manual.
FD: How significant is Trint, and how do you see it revolutionising other industries aside from journalism?
JK: I think it is a really big deal. When you see the traction we’re getting, the major news organisations, universities, corporations, governments, law courts, and healthcare systems, that are now working with us, you start to understand that this is a problem desperately needing a solution, and people see Trint as that solution. I was speaking to a producer of National Public Radio in Washington that we’ve just launched an enterprise pilot with, and she said to me: “I can’t imagine going back to life before Trint.” If you look at our Twitter channel you will see we get that feedback almost every day. It really is extraordinary. I feel like we’re sitting on the cusp of a cult, because when you solve a problem this easily and this accessibly, people really love you.
FD: That must be extremely rewarding, and something that inspires you to carry on?
JK: It’s really fun. I mean, this is hard, but succeeding and seeing it engage is unbelievably satisfying. People ask me all the time if I miss my old job. I was a really lucky guy, I travelled the entire globe, and saw a number of extraordinary things that most people only dream of seeing (and some that you don’t want to see, as a war correspondent). It was a challenging, adventurous, and educational career, and you couldn’t have asked for more. But I don’t miss it for a second. To reinvent yourself after 30 years, into career 2.0, and have a challenge this exciting and rewarding, and to see people engaging with it so passionately and enthusiastically – it’s just phenomenal. It’s so fulfilling.
Originally published here.