One year ago, open-source AI coding agent OpenCode debuted at a DevTools Toronto meetup with maybe 30 people in the room. That tool now has eight million monthly active users and expects to bring in around $25 million in annual revenue.
It’s all led by Jay V, a University of Waterloo alumnus who, clad in a sleeveless tank, looked more like a protein-powder founder than an open-source evangelist as he recounted his company’s meteoric rise on-stage at Toronto Tech Week’s Homecoming event a few weeks ago.
Jay V,
We were built on the premise that most of the world had still not discovered the value of agentic coding. We’re riding that wave now.
OpenCode
The world’s most popular open-source AI models, Qwen and DeepSeek, were both created in China. Meanwhile, US-based Anthropic hasn’t open-sourced any of its flagship models, OpenAI has only released a handful as open-source over the years, and Cohere has only just released its most powerful model as open source.
As you’d expect, making a product free results in greater adoption; 41 percent of AI models downloaded last year were built in China. Now, an agentic option straight out of Toronto has put its flag in the ground, allowing users to connect any model they want to get the job done.
BetaKit pulled Jay V aside shortly after he left the Homecoming stage, where he told us about the roots of his company’s open source ambitions, its challenges, and where he sees the industry going next.
The following interview has been edited for clarity and length.
Why open source?
Open source as a movement, and as a way to build software, is so underrated.
Everything that we use is dependent on it in some way, and it is really crazy to think that most of these things are done because somebody has some free time to put into creating a project that they really care about. All of the software infrastructure that we see in the world boils down to efforts like that. We’re not quite appreciating how crazy and how unique it is, and just being able to be a part of that and contribute feels good.
On the AI side, if AI is going to have the impact that it’s going to have in every single part of the economy, every single part of the globe, the diversity of these things really matters, and it matters that all of this is not concentrated in a couple of entities. Open source ends up being the best way to allow for that diversity.
What are the unique challenges of running an open-source business?
Initially, it is just getting over the hump that you’re just completely public. It’s not a very natural thing for a lot of people to do, especially when you’ve first graduated or you’re just getting started in your career.
The fact that you’re not that confident in yourself, and now, all of a sudden, anybody can just look at what you’re doing and comment on the mistakes that you’re making.
Cohere recently announced its own open-source model. Would you like to see more AI models open up?
For sure. I think the big problem here is that we’ve got OpenAI, Anthropic, and other model labs out there sucking up so much of the Western talent that there just isn’t enough to go around.
When we talk about open source, we’re effectively talking about all Chinese labs, and that’s obviously great in terms of diversity, because that’s something that you’d want, but at the same time, we’d want that here, both in the United States, in Canada, and in the West, in general. There’s a little bit of imbalance currently.
I think when you start planting the flag that we’re going to carry open source forward in the West, other people are probably going to follow.
You registered a patent for OpenCode recently. Why was locking down that IP important to you?
I think the name ended up mattering both in terms of what it symbolizes as being open source and being a coding agent.
People take a lot of liberty in building on top of our brand in some ways. Not that I think we care too strongly about exercising that, it’s just how valuable it is when thinking about building this out over the next 10 years. People want to support open source, and they hear OpenCode versus one of the other dozens of coding agents; it’s immediately obvious what OpenCode is.
If you were to name the most consequential decision you made that led to OpenCode’s fast adoption, what would it be?
There were different things along the way. When we had launched OpenCode, there weren’t a ton of open-source models for coding, we just bet that is a thing that was going to happen.
Right out of the gate, a big concern for us was making sure that we support every single model and provider, as opposed to how other people build in our industry, where they manually curate the list. We created this public repository separate from OpenCode that anybody could contribute to. If Cohere comes in, or any other company comes to open source, they don’t have to talk to us about it; they can just contribute to that repo, and it will show up in OpenCode and anywhere else that people are using it as well.
We’ve been in the agentic AI era for a minute now. Where do you think the next wave of AI adoption is going to come from?
I think what’s happening with the numbers that Anthropic is putting up in terms of revenue is getting people to over-rotate towards doing what they’re doing in enterprise. You’ve got OpenAI sitting on close to a billion active users, and even for them, when they see the numbers that Anthropic is putting up, they’re thinking, “oh, wait a minute, maybe the value actually is in the enterprise in the agentic world.”
RELATED: Canadian government gains access to Anthropic’s powerful Mythos model
I’m a little bit worried that we’re spending too much time thinking about the enterprise, and maybe not enough about what the average consumer wants, so I think that will probably be the next wave.
What’s next for OpenCode?
We were built on the premise that most of the world had still not discovered the value of agentic coding. We’re riding that wave now. Now, if most people have tried it and this is a part of their workflow, what are some obvious things that most of the early adopters are doing, but the rest of the world hasn’t quite done? Can we bake that into the product?
There are like a dozen apps on your phone, you probably just use one or two of them, and that’s what most of the world does. It’s just figuring out those key functions and then bringing them back to everybody.
BetaKit is the official media partner of Toronto Tech Week.
Feature image courtesy Toronto Tech Week 2026.