Five-year-old startup Mighty AI built a platform that uses input from everyday people to train computer vision systems, often on behalf of self-driving car companies. Today, the 74-employee startup has $27 million in funding from investors including GV, Alphabet's venture capital arm. In July 2017, Mighty AI was scaling quickly when 44-year-old CEO Matt Bencke received a devastating diagnosis. It thrust the company into a scary new world: one without Bencke at the helm. Co-founder and then-CTO Daryn Nakhuda tells the story of how he and the company dealt with, and eventually overcame, tragedy. --As told to Kevin J. Ryan

All four of us co-founders got recruited by Madrona, a venture capital firm here in Seattle that had developed the idea for Mighty AI. Three of us all knew each other from the local startup world. Matt was kind of a big-company guy; he had spent a decade each at Microsoft and Boeing. It made for a good balance of different kinds of experiences.

Matt was the kind of leader who would jump up on the table on Monday morning and give inspirational speeches with metaphors about climbing mountains and enjoying the views along the way. He was an incredibly fit guy. He rode his bike into work every day. For the month or so leading up to his diagnosis, he'd had some back problems. Everybody was just kind of laughing about it: "Well, he's finally getting old. I guess he is human after all."

He called me early one morning, earlier than we would usually talk. He said, "You know, something's wrong. I had to go to the ER last night. I'm in the hospital right now and they're running some tests." Twenty minutes later he called me again and said, "I've got bad news. I was diagnosed with pancreatic cancer. I've got to deal with this, so I'm going to step down, and I'd like you to be CEO." He had already talked to the board and made sure they were supportive of it.

I always wanted to be a CEO, but I wasn't expecting to drop right into a company with $20-plus million raised and 40 employees. It was like plunging right into the deep end.

In the early months there was always this big hole. At every meeting, we were waiting for that voice that wasn't there anymore.  

We all spent a lot of time talking about the situation and trying to support each other. We had just been getting some real traction as a company, so we were all very, very busy.

Even after Matt stepped down, we would talk with him very regularly. He would offer some ideas--things we should be doing and things we should be thinking about. He and I had a good, healthy relationship. He was helping me step into this new role.

Initially, we were all hoping that he would have a recovery at some point. After a few months, it became clear that wasn't going to be the case. He did as much as he could for the company before he became too sick to help.  

I was in Detroit, speaking at an event that Matt was supposed to speak at, when I got the news. At first my response was very reactionary: making calls and sending texts, telling the people you need to tell. Then, before I left Detroit that day, I got on one of those public rental bikes and rode along the riverwalk there, since Matt was an avid cyclist. That was when it really hit me. I thought about my friend Matt and his family. I thought about my own life, all the things I was looking forward to, and the reality that you never know what's waiting around the corner. And I thought about Mighty AI, and how it was on me to take the reins and make our collective vision a reality, without my partner by my side. 

Within the company, it was all about supporting each other. We told everybody: If you need to take a day off, if you want to grieve in your own way or grieve with your co-workers, we're here to support that. There were people who had been here for multiple years with Matt and people who had started after he had gotten ill. Everybody was going to have to deal with his death in their own way, and we wanted to give them room for that.

When Matt first got sick, we would ask ourselves, what can we do to help? Obviously we couldn't make him better. But we said the No. 1 thing we have control over is making this company a huge success, for all of us and for Matt and for his family. We said, "Let's make him proud. Let's make his family proud. Let's do something for his legacy."

Moving forward, one of the tougher things for me was trying to learn to be me and not trying to be Matt. He was a very outgoing, extroverted, inspirational leader. I was always the engineer, a little quieter, a little more reserved, a little less flowery in the way I talked. But that's one thing I think that has changed over time. There's certainly a need for inspiration, but I've learned I can do it in my own way. That means not trying to talk like Matt or reference things he would reference, but taking my own stories and figuring out how I can apply them and hopefully find a way to excite the team.

Matt lived a very big, well-rounded life, between his family and his music and his company. This was his first startup. It was of huge importance to him. He did so much for this company in the three years he was here. Whenever we talk about decisions we need to make as a company, we always come back to: What are our values and where do we stand? And most of those things were driven early on by Matt.

There's not a day that goes by without someone referencing him in one way or another. We have photos with him all around the office. We have a blood-orange tree in the lounge in his honor that we raised from a scrawny little sapling. I guess it's still pretty small.

Matt talked a lot about enjoying and valuing life. We have a company holiday at the end of July, around the day he stepped down. We let people take time off, do whatever they want. Last year some people went on a bike ride together. It's always a reminder for people to spend time on themselves, too. I think all of us look at life a little bit differently now.