After backing Zabidou at pre-seed, we’re doubling down in the seed round to support their vision of building low-cost, ultra-high-performance vision systems that bring quality control to the industries automation has always left behind: construction, fabrication, recycling, and the rest of the physical economy that keeps everything running.

We already knew what a gun operator looked like

We have long conviction in Cibby Pulikkaseril. Main Sequence backed Baraja, the solid-state LiDAR company Cibby co-founded and ran as CTO. He’s been candid that Baraja’s hardest lesson was spending years and capital chasing production contracts that were always just over the horizon, rather than revenue that was actually on the table. So when he came back to us with Zabidou, what struck was how deliberately he’d taken that lesson and rebuilt his entire operating model around it.

Cibby has a line he keeps coming back to: he goes where the money goes, he goes where the customers are, and he builds what the customers want. The company only builds for customers who are ready to pay today, not the ones who might be ready in five years. He’s permanently on a customer site, obsessively close to the people who want to use his product. To us, that’s the clearest tell there is that someone has actually internalised what it takes to build a scalable business, and it’s exactly the kind of second act we want to be part of: same technical brilliance, hard-won commercial discipline, a founder with something to prove.

The industries automation forgot

Spend any time talking to customers in physical AI and you notice something quickly: a huge share of the industrial world isn’t anywhere near “I want a robot.” Construction, fabrication, materials manufacturing; these are still, in 2026, at “I just want to know the quality of what I’m producing.” Day one of automation.

The reason isn’t lack of appetite. Existing vision systems can help, but they are usually too expensive, too difficult to deploy, or unreliable on the details that matter most in production.

As such, there is a big opportunity to build something low-cost enough that it doesn’t touch impede margins, high-quality enough to actually be useful, and scalable enough to go after a fragmented but enormous market that every capex-heavy incumbent has priced out. 

Zabidou is addressing just that with state-of-the-art vision, at roughly a tenth of the cost, with better accuracy and throughput than what’s currently on the market — cheaper, faster, better. 

Built by someone who actually understands the vision stack

The other half of our conviction is technical. Cibby is a world-class thinker on laser-based sensing; he knows exactly which parts of the vision stack matter for customers and which ones are R&D vanity projects. That shows up in how Zabidou has been built: a multi-modal sensor pod that runs in real time and in-line to give customers deep insights for applications like QA and at a cost point that opens the door to industrial scale. Zabidou’s AI agent-led forward deployment model also ensures that they can rapidly unlock value for their customers. 

It’s already showing up on the factory floor. In one construction-focused use case Zabidou’s system ingests engineering drawings, builds a live 3D reconstruction of parts as they move down the line, and flags any deviation back to both the floor and the engineering team. Today that job is done by hand across most of the industry, and the fragmented teams involved rarely see the same picture. The Tier 1s we spoke to confirmed it: alternatives exist, but they're expensive and imprecise on what matters most — surface quality, deformation, dimensional accuracy.

The customer pipeline tells the same story across construction, recycling and food manufacturing. This market has been underserved not because the demand is unproven but because nobody has been willing to build for its economics. Construction QA alone is a $20B+ market. 

What we’re backing this round

Cibby has spent the past eight months executing on his vision for the business: staying customer-obsessed, resisting the urge to over-build, and turning real pain points into real bookings. 

The next phase is about enabling deployment velocity across their growing customer base. We backed Zabidou at pre-seed because we believed a low-cost, high-performance vision system could unlock industries that had been shut out of automation on cost alone. 

Eight months on, the thesis is looking sharper. We’re glad to keep supporting Cibby and the team.

‍