Solo.io boss: I was wrong, I made mistakes – and that made me a better CEO
(2025/08/28)
- Reference: 1756369267
- News link: https://www.theregister.co.uk/2025/08/28/solo_io_interview/
- Source link:
Interview "I feel that a founder always needs to be a little bit stupidly optimistic." Solo.io CEO Idit Levine has been on an interesting journey in cloud computing since starting the networking and API management company in 2017.
If you're looking at a project that is successful today, it was after a big company leans into it and is actually paying people to work with you on that project ...
Solo.io is a cloud connectivity company. Its main cloud products are Gloo Gateway, a cloud-native API gateway built on Envoy, and Gloo Mesh, which aims to simplify Istio service mesh management for Kubernetes.
And, of course, there are the inevitable AI-related products. When we spoke to Levine, Solo.io had just announced an MCP gateway for kgateway (described by the company as "the ecosystem's most mature and widely deployed cloud-native API gateway"). Most recently, the company [1]announced that it had donated agentgateway to the Linux Foundation.
In 2021, the company [2]announced a $135 million funding round at a company valuation of $1 billion – a tenfold increase since the previous round in October 2020. It's impressive in the pre-AI era, but was taken, according to Levine, because she reckoned there was a bubble that was just about to pop.
"I predicted," she says, "that the bubble that was in 2021, all this crazy race, is going to disappear."
[3]
So, Solo.io took the money and bought Levine time to make some mistakes. "I made a lot of mistakes," she laughs. "I'm a first-time founder. I hired the wrong people in the leadership. I didn't understand what would fit in the company I wanted to build."
[4]
[5]
Levine reckoned that starting the company would be a case of surrounding herself with smart people and putting out a project. "In the beginning, we kind of put out some open source project. A lot of people here [she gestured at KubeCon] said, 'That's awesome!'
"It's rewarding. We will be very famous and everybody will adopt it. You will think that you put a good project out there – that's what I was thinking – it would be like Docker. Everyone would see it was good. No one will compete. And then we will create the next Docker. Oh boy, was I wrong."
[6]
Levine found herself in a world of conflict and politics. While she doesn't regard herself as a particularly good player, she did learn to be more sensitive.
"I made mistakes which hurt the company," she admits, and embarked on a restructure. "I basically moved division by division in the company and redid it, starting with sales."
The process has, by all accounts, been a success, even as Levine dealt with personal issues while working her way through the organization to unpick her earlier missteps. "My point," she says, "is that it's not like on the outside – you succeed – it's not. Oh my God, there's so much hard work, and so much hard work that's not related to the technology!
[7]
"I've learned so much, and I'm a way better CEO because of it."
Levine started Solo.io when Kubernetes was still in its relative infancy. "When we started Solo," she recalls, "Kubernetes was already built. There was already CI/CD, GitOps, and so on. So I said, 'OK, what is the next problem?'
"I chose networking because I believed that that will be something that (a) no one addressed yet, and (b) it's a big problem, right?"
Levine is correct. Networks were a pain point in Kubernetes, often requiring careful configuration to make them work. "I kind of feel like we did so much like we messed up, I'll be honest, on the networking," admits Levine. "Not only us, all the community."
[8]The air is hissing out of the overinflated AI balloon
[9]One long sentence is all it takes to make LLMs misbehave
[10]CIO made a dangerous mistake and ordered his security team to implement it
[11]VMware before Broadcom was 'A unicorn in fluffy cloudland'
"We made it so complex to use that people said the trade-off is not working... so what we did on the last release is basically fix it. I think, right now, we nailed it. I can go line by line in the cloud and show you!"
Levine reckons the same approach will pay dividends in the world of AI. "When we go back to agentic, the first thing we need to invent is orchestration... so that's where we started and that's what we did with kagents."
The tech chief's advice for other startups is simple. Solving a perceived problem is part of the work, but getting paying customers on board is another, as is listening to those customers. "Honestly, when we open sourced the first project, Gloo, it was very far away from what it is today." And the reason is requests from paying customers: "Oh, that's great, but we want this and we want that, and they help us build... we learn enterprise through them."
Levine's second tip is an acknowledgment that people need to be paid. Relying on the generosity of people with their free time does not necessarily equate to a successful project. She says: "What I do believe is happening is that if you're looking at a project that is successful today, it was after a big company leans into it and is actually paying people to work with you on that project."
A project does not necessarily succeed because it's better, but "it's because everybody is leaning into it and then they make it better," according to Levine.
We spoke to Levine at KubeCon 2025, where it appeared almost every exhibitor had slapped an AI suffix on their product. "Someone said, 'Well, AI today... You have to put AI in the title in order to get into this conference,'" she says.
"And someone from CNCF said, 'That's not true... if you look really closely, you'll discover that multi-cluster is the subject,' and that's interesting.
"To me, it's like, 'Oh lucky us!'" laughs Levine. "We saw it. We were smart to predict it."
But that's not the point. Her advice to listen to customers and to own mistakes, coupled with some canny timing around accepting investment, are useful tips to take on board when pondering how to go from startup to a billion-dollar valuation. ®
Get our [12]Tech Resources
[1] https://www.solo.io/blog/solo-contributes-agentgateway-linux-foundation
[2] https://www.solo.io/press-releases/solo-io-advances-application-networking-market-with-135-million-series-c-funding
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[8] https://www.theregister.com/2025/08/25/overinflated_ai_balloon/
[9] https://www.theregister.com/2025/08/26/breaking_llms_for_fun/
[10] https://www.theregister.com/2025/08/25/who_me/
[11] https://www.theregister.com/2025/08/25/yves_sandfort_comdivision_vmware_interview/
[12] https://whitepapers.theregister.com/
If you're looking at a project that is successful today, it was after a big company leans into it and is actually paying people to work with you on that project ...
Solo.io is a cloud connectivity company. Its main cloud products are Gloo Gateway, a cloud-native API gateway built on Envoy, and Gloo Mesh, which aims to simplify Istio service mesh management for Kubernetes.
And, of course, there are the inevitable AI-related products. When we spoke to Levine, Solo.io had just announced an MCP gateway for kgateway (described by the company as "the ecosystem's most mature and widely deployed cloud-native API gateway"). Most recently, the company [1]announced that it had donated agentgateway to the Linux Foundation.
In 2021, the company [2]announced a $135 million funding round at a company valuation of $1 billion – a tenfold increase since the previous round in October 2020. It's impressive in the pre-AI era, but was taken, according to Levine, because she reckoned there was a bubble that was just about to pop.
"I predicted," she says, "that the bubble that was in 2021, all this crazy race, is going to disappear."
[3]
So, Solo.io took the money and bought Levine time to make some mistakes. "I made a lot of mistakes," she laughs. "I'm a first-time founder. I hired the wrong people in the leadership. I didn't understand what would fit in the company I wanted to build."
[4]
[5]
Levine reckoned that starting the company would be a case of surrounding herself with smart people and putting out a project. "In the beginning, we kind of put out some open source project. A lot of people here [she gestured at KubeCon] said, 'That's awesome!'
"It's rewarding. We will be very famous and everybody will adopt it. You will think that you put a good project out there – that's what I was thinking – it would be like Docker. Everyone would see it was good. No one will compete. And then we will create the next Docker. Oh boy, was I wrong."
[6]
Levine found herself in a world of conflict and politics. While she doesn't regard herself as a particularly good player, she did learn to be more sensitive.
"I made mistakes which hurt the company," she admits, and embarked on a restructure. "I basically moved division by division in the company and redid it, starting with sales."
The process has, by all accounts, been a success, even as Levine dealt with personal issues while working her way through the organization to unpick her earlier missteps. "My point," she says, "is that it's not like on the outside – you succeed – it's not. Oh my God, there's so much hard work, and so much hard work that's not related to the technology!
[7]
"I've learned so much, and I'm a way better CEO because of it."
Levine started Solo.io when Kubernetes was still in its relative infancy. "When we started Solo," she recalls, "Kubernetes was already built. There was already CI/CD, GitOps, and so on. So I said, 'OK, what is the next problem?'
"I chose networking because I believed that that will be something that (a) no one addressed yet, and (b) it's a big problem, right?"
Levine is correct. Networks were a pain point in Kubernetes, often requiring careful configuration to make them work. "I kind of feel like we did so much like we messed up, I'll be honest, on the networking," admits Levine. "Not only us, all the community."
[8]The air is hissing out of the overinflated AI balloon
[9]One long sentence is all it takes to make LLMs misbehave
[10]CIO made a dangerous mistake and ordered his security team to implement it
[11]VMware before Broadcom was 'A unicorn in fluffy cloudland'
"We made it so complex to use that people said the trade-off is not working... so what we did on the last release is basically fix it. I think, right now, we nailed it. I can go line by line in the cloud and show you!"
Levine reckons the same approach will pay dividends in the world of AI. "When we go back to agentic, the first thing we need to invent is orchestration... so that's where we started and that's what we did with kagents."
The tech chief's advice for other startups is simple. Solving a perceived problem is part of the work, but getting paying customers on board is another, as is listening to those customers. "Honestly, when we open sourced the first project, Gloo, it was very far away from what it is today." And the reason is requests from paying customers: "Oh, that's great, but we want this and we want that, and they help us build... we learn enterprise through them."
Levine's second tip is an acknowledgment that people need to be paid. Relying on the generosity of people with their free time does not necessarily equate to a successful project. She says: "What I do believe is happening is that if you're looking at a project that is successful today, it was after a big company leans into it and is actually paying people to work with you on that project."
A project does not necessarily succeed because it's better, but "it's because everybody is leaning into it and then they make it better," according to Levine.
We spoke to Levine at KubeCon 2025, where it appeared almost every exhibitor had slapped an AI suffix on their product. "Someone said, 'Well, AI today... You have to put AI in the title in order to get into this conference,'" she says.
"And someone from CNCF said, 'That's not true... if you look really closely, you'll discover that multi-cluster is the subject,' and that's interesting.
"To me, it's like, 'Oh lucky us!'" laughs Levine. "We saw it. We were smart to predict it."
But that's not the point. Her advice to listen to customers and to own mistakes, coupled with some canny timing around accepting investment, are useful tips to take on board when pondering how to go from startup to a billion-dollar valuation. ®
Get our [12]Tech Resources
[1] https://www.solo.io/blog/solo-contributes-agentgateway-linux-foundation
[2] https://www.solo.io/press-releases/solo-io-advances-application-networking-market-with-135-million-series-c-funding
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_specialfeatures/cloudinfrastructuremonth&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aLAot9yrcYQB0dTHxTfTKwAAAI4&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[8] https://www.theregister.com/2025/08/25/overinflated_ai_balloon/
[9] https://www.theregister.com/2025/08/26/breaking_llms_for_fun/
[10] https://www.theregister.com/2025/08/25/who_me/
[11] https://www.theregister.com/2025/08/25/yves_sandfort_comdivision_vmware_interview/
[12] https://whitepapers.theregister.com/
Refreshing
SVD_NL
Very refreshing to see a CEO with a healthy attitude. These days, especially in tech/software/AI startups there's too many lead figures defined by their ego.
Good to see her admit her mistakes, acknowledge different contributions, and staying true to the path she's set out on, without letting the countless people tugging her arm win.
Anonymous Coward
I've done a few technical due diligence gigs prior to PE buy-outs of smallish tech companies. In my experience the ideas people who started them are not always the best people to turn them from a start-up to a functioning, growing company. In one case the founder was up there with the brightest people in the world but his butterfly brain wasn't capable of dealing with the day-to-day neccessities of running a business with a few dozen people who needed paying every month. The company was on the bones of its arse and lucky to be bought by a PE before it collapsed. He wanted to stay on as CEO after the buy-out but the PE were very firm and, on top of the hefty pay off he got from selling his company, they gave him a part-time role as director of innovation on the understanding that he'd stay out of the day-to-day operations.
SVD_NL
The ideal situation is having one person who is the technical/creative mastermind, and partner them with someone who understands how to grow and run a healthy business. They rarely come in one and the same package!
Whatever you think about Solo.io, you have to admit it takes guts to be a female CEO and it takes even more guts to freely admit that you made mistakes.
I'd like to see more male CEOs admit that, eh Léo Apotheker ?