Oliver Keh on Pivoting, Clinical Trials, and Real Product-Market Pull
Episode #1
About the Guest
Oliver Keh is the founder and CEO of Gleam (YC S23), an AI copilot for clinical research sites that automates the hours of data entry that is typically done manually by site staff every week. He previously built healthcare software at Amazon Care and worked at Bolt. Oliver holds a B.A. in Computer Science from Hamilton College and is based in San Francisco.
Editor’s Note:
I first met Oliver Keh, founder and CEO of Gleam, in the summer of 2025. His down-to-earth personality and always warm smile made our office coffee breaks a delight. Behind the gentleness of his personality is the relentless resilience that brought him here. Before founding Gleam, Oliver spent years building software at Amazon Care, Bolt, and earlier-stage startups. Like many founders, his journey has been anything but linear.
After raising a pre-seed round and joining Y Combinator, Oliver quickly realized his first idea wasn’t working. What followed was more than a year of persistence, sleepless nights, pivots, and deep reflection. Then came the breakthrough: a viral LinkedIn video about clinical-trial data-entry inefficiencies that lit up the industry and pulled Gleam into the spotlight. Today, Gleam is building an AI copilot for clinical research sites that automates critical operations work, like data entry, that is being done manually across the industry.
In this candid conversation, Oliver shares what it feels like to pivot under pressure, how transparency shaped his team culture, and why history, not just tech, guides his leadership.
Interview:
Xiao He (Mother of Success):
Oliver, I've known you for a while, but there are many things I would love to learn more about you. Gleam right now is doing really well. You are in clinical trials data entry and signing amazing contracts, congrats on that. I just remember, maybe just a few months ago, you were going through a lot, and then suddenly a video you posted gets really popular and everything just opened up. So I think your experience would be really inspiring to a lot of the readers at Mother of Success. But before that, could you introduce yourself to our readers?
Oliver Keh (Gleam):
Sure, I'm Oliver, founder and CEO of Gleam. Gleam is building an AI-powered eSource to EDC data connector, a very complicated term that most people do not understand, but in software engineering terms you can think of it as an integration-less ETL that takes data from one system where clinical researchers document trial data and puts it into a system that's owned by pharmaceutical companies.
My background is in healthcare. I spent basically my whole career working as a software engineer on technical systems in the healthcare industry. I've worked on electronic medical records. I've worked on building tools for clinicians and patients all at Amazon Care, which is Amazon's first push into healthcare delivery. That was an awesome experience. I got to see it from some of the earliest days, back when it was still a confidential special project within AWS, and then getting to see it roll out, publicly announced, and actually getting to use it as a customer.
And then I briefly transitioned out of Amazon. I decided I wanted to go experience the startup world. Amazon Care did feel like a startup, but it obviously had the guardrails and blank check of AWS. So I wanted to see what it was like outside of that. I briefly went to another startup for a few months, but realized that I wanted to be at an even earlier stage than that startup was. And so I had to go through a mental calculus of, do I want to go and be an early hire at a startup, or do I want to start my own. Both have pros and cons, but based on where I was at that time, I thought, you know what, I might as well start a company.
I had a few ideas that I wanted to play with, and luckily I had made enough savings at the time to be comfortable enough to take a risk like that. So I did, and I spent a few months playing with some ideas, landed on one thing that had some promise, raised Gleam’s pre seed round, and then got into Y Combinator. The company was kind of in a unique position where we had raised money before Y Combinator.
Xiao:
You raised pre seed money and got YC. What was your idea that made you into YC?
Oliver:
At the time, I mentioned I spent a lot of time in the healthcare industry, so obviously a lot of the ideas that I was exploring were in the healthcare space. I looked at building software for private medical clinics, hospital systems, but found those sales cycles to be really brutal. I just couldn't find anything in particular that resonated enough with folks to drive hyper growth.
So I started moving around a little bit, and I started talking to companies about their health benefits, what they're offering to employees in terms of insurance. I learned about how complicated and expensive health insurance is for small businesses, and consistently found that to be a pretty big pain point for folks.
I ended up looking at, could we build a software system that would help simplify the model of health insurance that larger companies use, and bring it to smaller businesses. The idea being self funded health insurance. And so that was the idea that got into Y Combinator.
I quickly realized that it wasn't really going to work for many different reasons. So then I pivoted into the HRA space, which is a kind of self-funded model, not truly a self funded health plan, but basically it's a way to offer money to employees to spend on healthcare services and things like that.
Xiao:
Quick follow-up question, you mentioned that the previous idea, you quickly realized it didn't work. Just to get the timeline, you were in the Summer 23 batch, and you got into YC with an idea. When did you realize the idea that you got in YC with didn't work, and why it didn't work?
Oliver:
It was probably within a month or two. At that time the YC batch was a little longer than it is today. It was three months versus now, I think it's about two months. I would say it was within two months that I realized, okay, I think we need to find something different.
Xiao:
That realization is actually pretty fast. In a way, it's really good that you realized it that fast.
Oliver:
Yeah, I think it was mostly that people were interested in the product, and so they were willing to talk. And they were buying into the idea. The problem is with the execution. With this kind of model, it's very, very difficult to sell a health insurance product, I learned. Because as a company, you're really only reevaluating your health insurance benefits once a year. Any changes to your health benefits are a huge undertaking for most companies.
So what I found was that when selling such a complicated insurance model, people quickly got overwhelmed with how complicated it was, and the sales cycle wasn't converting as much. I could get people through the earlier stages of the funnel, but then they would drop off.
Xiao:
After you realized it didn't work, what did you do? Did you pivot a little bit, or did you pivot completely?
Oliver:
I pivoted a little bit. I mentioned that I pivoted into HRA benefits. It's a simpler self-insurance model. Doesn't require an employer to switch their entire insurance package. It's easier for people to buy into. You can think of it as a set amount of money that an employer gives employees to spend on healthcare services.
For employers, it's great because most folks don't end up using their healthcare benefits. So buying everybody a really expensive insurance plan is usually not a great use of money. It's easier to have people on a Silver plan, which is kind of a mid-tier plan, and then augmenting that with an HRA, X amount of dollars to make it feel like a platinum plan. People really liked that, and so it ended up selling pretty well. We got a good number of customers on that.
But again, over the course of another year, that I really dug into that, we still ran into some of the challenges I mentioned with the self-funded insurance model. It's really hard on an insurance product to get people to make changes once they already have a model in place.
And another challenge selling to startups was that as we were adding new customers, our older customers were shutting down. Most startups don't succeed, so new customers we added were constantly having to replace lost revenue. And there wasn't a good enough channel that I discovered to become the default insurance product for startups. I wasn't able to crack going up market either.
So it was about a year, maybe a little over, that I decided, you know what, it's not growing as fast as we want. This is not going to be a huge company. Let's do something else.
Xiao:
I want to dig a little bit deeper into this experience. Oliver, how were you feeling at that time, toward the end when you were deciding to pivot?
Oliver:
Yeah, I think any startup founder will tell you that when it comes to the decision point, or leading up to it, it's a really stressful time. Because you invested so much time and sweat and blood and tears into this idea. There were a lot of sleepless nights and a lot of effort that went into trying to make it work.
And honestly, I think a lot of startups think that putting more hours into an idea that's not going to work will make it work. I think you do have to work extremely hard, so I'm not saying the right way to do a startup is taking it easy, because it definitely is not.
I just think you have to know when to cut your losses and say, this schedule I'm working so hard at could be better spent on something else that I really think is going to have an outsized outcome. That was the realization I came to. We were working so hard on this thing, and initially I had full conviction it was going to work, but over time that conviction eroded.
We had all this money in the bank because we were really diligent. We didn’t really burn much. So we had the flexibility to say, let's just kill that thing and go do something else.
Xiao:
Great to hear you managed your financials really well so you had the full runway when you were deciding to pivot, instead of running out on an idea that didn’t work. I’m curious to learn more about the moment you decided to pivot. Was there any particular event or trigger that made you really decide to?
Oliver:
No, I don’t think there was one particular moment. It was probably a series of moments, where we got so close and then it fell apart. Or initiatives I tried with the team to generate more traction that didn’t work out. Or working so hard on product releases that didn’t get used. A lot of those things kind of added up over time.
Xiao:
I know when I met you at the coworking space here in San Francisco, Daren was on your team. Was he with you before you pivoted or after?
Oliver:
Yeah, so there were three of us at the time. Daren was our intern. He’s coming back next summer for his full-time role. And we have Alan, who’s our founding engineer, based in Washington DC. They’ve both been on the team for over a year now. And so, yeah, they went through the whole pivot process alongside me.
Xiao:
Were you transparent with them, like what’s working and what’s not working? Or did you try to hold up all the stress and let the team relax?
Oliver:
I tried to be very transparent with the team, and I really tried to include them as much as possible in the pivoting process. There were weeks where we just spent time researching particular ideas or industries, interviewing folks. The team spent a lot of time building prototypes, and I was out doing customer interviews and whatever else.
It was a good two, three months of grinding to find something that really had a lot of potential. Before I took us through the pivoting process, I had a whole framework for how I wanted it to go. I reflected a lot on what worked and what didn’t work, and what I thought was a better way to approach building a startup and something that people cared about.
It paid off. That thesis and that whole reflection process was probably one of the most valuable things I did leading up to the pivot. Because at the point we’re at now, we’re just blazing ahead.
Xiao:
You’re doing so well! Just a few weeks ago, I told you it was my most difficult time at RegHero. I just couldn’t sleep. And I asked you, and you said you couldn’t sleep before pivoting, but afterwards you could sleep soundly. That’s awesome. I’m curious to learn more about the two or three months of research. How many ideas, or what types of ideas, did you evaluate, and what was the process like when you were evaluating them?
Oliver:
Yeah. Talking about the sleep component, definitely. I remember watching an interview with Jeff Bezos where he said something about how anxiety and stress are very similar. But anxiety is when you don’t know what to do, and stress is when you actually do know what to do, you just have to go and do it.
I was feeling a lot of anxiety leading up to this pivot and decision, and that’s what was keeping me up at night. As soon as I decided, hey, we need to pivot and do something else, that turned into stress, and then I was able to sleep at night. That was great.
As for the process, I think the shortest time I spent on an idea was like a day, just exploring something. The longest was maybe three weeks. A lot of that time was spent learning. I decided we needed to do a 180 pivot, leave everything behind, start from scratch.
Luckily, all of our investors have been incredibly supportive through it. They lent a lot of their expertise and knowledge and helped me think through what areas I should consider. We have dozens of Notion pages documenting learnings, thoughts, everything. It was a lot of time spent just trying to find something.
And then we found out about clinical trials. It happened to be pretty close to healthcare. Not intentional, kind of coincidental, but luckily, my background working on electronic medical records and being a software engineer led to a lot of key insights about how to build the core product and how to solve this problem that so many companies have tried to solve for years and haven’t been able to. It felt kind of like magic.
Xiao:
I think it’s magic, but also all the years and effort you put in before Gleam, like your experience at Amazon Care and your solid software engineering background. It zigzagged, but it eventually led you here. I’m curious about that video that went viral. Was it after you decided you wanted to pivot into clinical trials? Did you invest more time into making that video, or was it more like testing things out?
Oliver:
At that point, I had realized there was something here, and I had a very clear idea of how we were going to solve this problem. We talked to dozens of people in the industry, from ground-level research coordinators who are the ones actually running the clinical trials, all the way to pharma executives and CROs, to really get a full picture of why this problem exists.
What I didn’t have at that point was, did people want to buy this product. So before we invested more time, I wanted to get an idea of what the market reception would be. How serious are people about buying this product.
It turns out people are very serious. That video really blew up. The traction it got, the sales inbound we got from everyone from independent research clinics to Fortune 100 pharma companies, it was overwhelming. And it’s still carrying through. People are still talking about Gleam, and we’re still getting more leads from those few impactful moments.
Xiao:
That reminds me of Rob Snyder’s “PULL” framework. If there really is a demand, and it’s a real pain point, people will pull the product out of you. Oliver, your video, from just posting it on social media, got so much traction and inbound sales inquiries. That’s amazing!
Oliver:
Oh yeah, we’ve got a lot of them. Typically, pharmaceutical deals work on the order of 18 months. But what was really compelling was that a lot of these companies were reaching out and talking about timelines of months, like two to three months, which is unheard of in pharma. To be able to sign deals like this, it’s been quite amazing. Like you said, it feels like people are pulling the product out of us.
Xiao:
Resilient, persistent, and you never give up. You pivoted a little when you first got into YC, and then you pivoted 180 degrees. And all the experiences you had before just poured in this direction. That’s amazing. I’m really happy for you. What’s next for Gleam?
Oliver:
The world is our oyster. We’re growing the team. We’re looking to hire more software engineers here in the Bay Area. We just hired Daniel Perez as our head of clinical strategy and partnerships. He’s got 16 years of experience operating clinical trials at every level. From minute one of meeting him, I knew he had to be part of the team.
We’ll continue to make big product improvements, roll out the product to more research sites and pharma sponsors, and expand far beyond just EDC data entry.
Xiao:
One last closing question I plan to ask every interviewee at Mother of Success. What is one book or podcast you would recommend to founders and aspiring founders? It doesn’t have to be about startups.
Oliver:
In terms of podcasts, How to Take Over the World is an amazing podcast. I listen to it every time a new episode comes out. It goes into great detail about the lives of some of the greatest leaders of history. One of my favorite series of episodes was about Hannibal Barca, who famously crossed over the Alps with an army from Carthage. Through sheer ingenuity and creativity, he was able to take on and almost defeat such a powerful force, which was Rome.
There are many great leaders that I find so inspiring. And they’ve taught me a lot about how to think about building a company. I think there’s so much to learn from studying not just great tech founders, but great leaders from history as well.
Xiao:
Absolutely. Awesome. Thank you so much, Oliver.
About Mother of Success
Mother of Success is a founder-led publication sharing practical, hands-on guides and candid interviews. Our mission is simple: to turn real wins and hard lessons into actionable insights for your next build.