From Internal Tools to Product: How Voxel51 Discovered Their Real Value Proposition
Sometimes your most valuable product is the one you build to solve your own problems. In a recent episode of Category Visionaries, Jason Corso revealed how Voxel51’s journey from AI consulting to infrastructure started with a crucial realization: their internal developer tools were more valuable than their consulting solutions.
The Consulting Trap
In 2016, armed with deep expertise in computer vision and a grant from NIST, Voxel51’s founders thought they had it figured out. “Our thought was like, well, we know how to understand video. Let’s just build a system in which our users, our ultimate customers down the line, could upload their video to our SaaS offering and we’ll find them the insights in the video,” Jason explains.
The early signs were promising. They landed pilot projects with the Baltimore City Police Department and clients in automotive and insurance. But a pattern emerged that would force them to rethink everything.
“For every vertical we went into, we had to spend months getting data, labeling the data, like training models on it, and then were able to get insights,” Jason recalls. “It really didn’t generalize that well.” The consulting model was turning them into a services business, requiring custom work for each new client.
The Hidden Product
While building these custom solutions, the team had developed internal tools to manage their development workflow. These tools weren’t the product they were selling – they were just the infrastructure they needed to do their work effectively.
But as they struggled with the consulting model, they realized something crucial. The problems they were solving internally – managing data, evaluating models, understanding failure modes – were universal challenges in AI development.
The Pivot to Infrastructure
Rather than continuing to push their consulting services, they made a bold decision: “We ultimately pivoted way earlier in the lifecycle to kind of meet the users where they were at and release the developer tools that we had been building for that consulting work.”
This wasn’t just a product pivot – it was a complete rethinking of their go-to-market strategy. Instead of trying to sell finished AI solutions, they would provide the infrastructure that AI developers needed to build their own solutions.
Building Trust Through Authenticity
The transition wasn’t immediate. As Jason notes, “One, it was a little too early to understand and spend money on video as a value prop and the ultimate lifecycle of businesses.” But they stayed committed to building authentic value.
“We never talk about vaporware. We never promise things in our customer discussions,” Jason emphasizes. “We were very careful to build that trust in that relationship over time.” This commitment to authenticity extended to their product development and marketing approach.
The Power of Open Source
In 2020, they made another crucial decision: releasing their core technology as open source. “We give away the full machine learning stack, the software, as long as it is one user, one machine, local data,” Jason explains. This move helped them build credibility with developers and understand how their tools were actually being used.
Lessons for Technical Founders
Voxel51’s journey offers crucial lessons for technical founders. First, sometimes your most valuable product isn’t what you’re trying to sell – it’s the tools you build to solve your own problems. Second, meeting users where they are often means providing infrastructure rather than finished solutions.
As Jason puts it, “It’s not like a sexy thing like data set quality or whatever, but we think of it as critical. Like roadways in American cities are not sexy either, but you need them to get around.”
For founders navigating similar pivots, the key is recognizing when your internal tools might actually be your real product. Sometimes the best opportunity isn’t in building the final solution, but in providing the infrastructure that helps others build those solutions more effectively.