When Insurance Channels Create Systematic Distrust: Voxel's Direct Enterprise Pivot
Vernon O'Donnell inherited a company with Carnegie Mellon-trained AI engineers, proven computer vision technology, and a $100-180 billion workplace safety market. The technology worked. The go-to-market didn't.
Two years into his tenure as Voxel's CEO, Vernon has transformed the company from a broken insurance-led channel strategy to direct enterprise sales with five Fortune 50 customers. In a recent episode of BUILDERS, he shared the specific pivots that drove 50% customer expansion rates—and why he believes the entire concept of technical moats died around 2019.
Why Insurance Carriers Make Terrible Channel Partners
Voxel's founding team pursued insurance carrier partnerships to identify high-risk customers. The logic: carriers know which companies have expensive workplace injury claims, creating a built-in qualification mechanism.
The execution revealed a structural flaw.
"The issue is you're already coming from a place of distrust," Vernon explained. "No one wants to share data back to insurance in an unfettered fashion from a carrier perspective."
Companies won't share the operational footage and safety data Voxel needs because they assume it flows back to insurers—potentially impacting premiums or coverage. The channel partner becomes the barrier.
Vernon killed the insurance channel entirely. "We pivoted away from a channel first and partner first approach. We kept really positive relationship with the carriers, insurance side and the brokers. But we built our own direct motion."
Insurance relationships didn't disappear—they transformed into validation mechanisms. Carriers now confirm ROI and provide credibility without controlling access.
The lesson for B2B founders: channel strategies only accelerate when partners genuinely enhance trust. When they create systematic friction, direct motion beats perfect targeting.
Deal Complexity Creates Natural Market Floors
Vernon faced the classic mid-market versus enterprise decision. He chose enterprise based on a single insight: buying complexity doesn't scale with customer size.
"The complexity of buying AI doesn't change mid market to enterprise," Vernon said. "And so there's really no inhibiting factor to go and going to that end market."
AI purchases require technical validation, multi-stakeholder buy-in, security reviews, and procurement cycles regardless of company size. If mid-market deals take six months with $100K ACVs and enterprise deals take six months with $500K ACVs, the ROI calculation favors enterprise.
This inverts conventional wisdom about moving upmarket. Most founders assume they need to "earn" enterprise readiness through mid-market success. Vernon's framework: if your product's inherent complexity already demands enterprise-length sales cycles, you're paying the enterprise cost without capturing enterprise value.
Voxel moved upmarket immediately and now works with five of the Fortune 50.
Fourteen Days From Kickoff to Live Results
The most impactful transformation came from redirecting engineering resources.
Early Voxel focused on expanding AI pattern recognition—more workplace hazards detected, more risk categories, more impressive demos. The technical team was proliferating capabilities.
Vernon shifted the entire focus to implementation speed. "We put a lot more of our resourcing from a technical perspective around just friction reduction."
The specific outcome: "Now we're typically live in a customer from the kickoff call to actual results in about 14 to 20 days. That's the algorithm learning the space, that's the technical provisioning, the end users seeing the information and us providing recommendations."
Fourteen to twenty days includes camera provisioning, algorithm training on the customer's specific environment, user onboarding, and delivering actionable safety recommendations. Not a demo. Live operational value.
This "excellence in the thing that we do well" drove nearly 50% of customers to expand their deployments.
The compounding effect: fast implementations create internal champions before organizational skepticism builds. Slow rollouts—regardless of feature depth—give detractors time to question ROI and stall expansion.
Technical Moats Died With VLMs and AI Coding Tools
Vernon's most provocative argument challenges B2B software's core assumption about defensibility.
"I don't fundamentally believe in technical moats anymore. I think that is 2018 thinking," he said. "You went on customer value, you went on speed to value, you went on retained value, repeated value. It's not about the tech only."
His thesis: video language models, AI-assisted coding, and accessible compute flattened technical differentiation. Capabilities that took specialized teams years to build now take months—or get commoditized by foundation model providers.
"You have to have a good tech team. Don't get me wrong, you have to have really good engineers," Vernon clarified. "But if you're going to sit back and say we're going to build the best technical product and the rest will come from there, you just are going to fail."
The shift challenges how founders think about R&D allocation. Technical excellence remains necessary for baseline credibility. But it no longer creates multi-year competitive advantages the way it did pre-2019.
The new defensibility: speed to demonstrable customer outcomes, retained value through continuous delivery, and distribution capability at scale.
Extract Problems, Not Solutions, From Customer Feedback
Vernon confronts founders who cite Henry Ford's "faster horse" quote to justify ignoring customers.
"Those things are what arrogant founders tell themselves to justify not listening," he said. "Henry Ford 100% listened to the market. He just synthesized it to a different application."
His reframe: customers told Ford they needed point A to point B faster. Ford heard the underlying problem—speed—rather than the literal solution. The discipline is extracting core needs from poorly articulated requests.
Vernon learned this lesson through a costly mistake in a previous role building sports technology. "I let that frustration take control," he recalled about a difficult international customer. "Were they a prickly group to work with? Sure. But ultimately, if you took the emotion out and listened to what they were actually saying, it would have made the product better."
He let tone obscure valuable product feedback. "That would have been a great product enhancement that would have benefited all of our customers."
The framework: separate emotional delivery from informational content. Prickly customers often signal real product gaps—they're just bad at articulating them constructively.
Build Distribution Before Winning The Account
Voxel's Fortune 50 retail customers operate thousands of locations. Vernon recognized that signing the contract meant nothing without deployment infrastructure.
"God forbid you win an account like that. And then they say this product is great, let's hit go. And then you're stuck without the proper hardware, without the vendors upstream, without the ecosystem to deliver. That's how you lose those accounts."
Voxel built distribution capability—hardware vendor relationships, deployment processes, installation teams—before closing deals requiring it. The alternative: win strategic logos you can't implement, poisoning expansion and creating reference-ability problems.
This inverts typical startup sequencing. Most companies build distribution infrastructure after proving they can win accounts. Vernon argues that for physical deployment businesses, that sequence guarantees implementation failures that kill momentum.
The framework: if closing a strategic account creates operational requirements you can't currently meet, build the capability before aggressive sales pursuit. Logo wins without successful delivery destroy more value than they create.
Moving Fast On Talent Misalignment
Vernon's transformation required significant talent changes, particularly in go-to-market roles. His approach: move decisively while treating people with dignity.
"We had a lot of talent transitions. I also am a big believer of like, we're not going to quibble over severance, meaning we want to pay people a fair separation. We want to make this something where they leave with grace and dignity."
The logic: "Why quibble over the margins when you have a bigger problem to solve from a transformation perspective."
Enterprise sellers require different skills than partner-channel sellers. Once the motion changes, talent misalignment doesn't self-correct through coaching or time. The severance cost is negligible compared to months of continued underperformance during critical transformation periods.
Vernon's discipline: "Once you know something's not going to work, it's not going to work. Like you don't like hoping against hope or praying that it's going to turn around is a waste of everybody's time and resources and just an impediment to growth."
Pay fairly, move immediately, redirect focus to building the new motion rather than managing out the old one.
The Industrial Intelligence Vision
Vernon sees Voxel's safety platform as an entry point to broader industrial intelligence. Workplace safety issues correlate with operational inefficiencies—poor warehouse layouts, suboptimal workflows, staffing problems.
"How do you start to build this out to where it's not just about saying you need a lift table here, but actually how do you design the warehouse that you don't even have to contemplate the notion of a lift table," Vernon explained.
The vision: use computer vision to optimize entire industrial operations, particularly as robotics and automation increase. "What those companies are lacking, and I think this is an inhibitor to success, they know they need automation. They know they need more robotics... Where and how. Where is the highest yield? How do they get the most benefit?"
Companies investing in industrial automation lack context on optimal placement and application. Voxel's comprehensive view of operations—currently focused on safety—positions them to answer those questions.
Vernon's transformation of Voxel demonstrates that fixing broken go-to-market isn't about optimization—it's about honest assessment of structural problems and decisive action. Insurance channels that create distrust don't improve with better messaging. Technical features don't compensate for slow implementations. And technical moats, in 2026, primarily exist in founder pitch decks rather than sustainable competitive advantages.