Why I’m Building Again

From Turnarounds to Vertical AI

I didn’t expect to start another company.

After years as a serial entrepreneur, and helping turn around Clearco, I wasn’t convinced I wanted to jump back into the founder seat. I know the highs of hypergrowth, but I also know the cost of burning years trying to find product-market fit. I also genuinely enjoyed running efficient organizations with strong operators, clear accountability, and high execution quality.

I was toying with the idea of specializing in turnarounds: larger teams and budgets, but also complex problems to unravel. It would satisfy my need to create order out of chaos while still having real resources.

AI came to rock this idea. I’ve been experimenting with AI since DALL·E had a waitlist, with plenty of fun builds and hallucinations (thanks, ChatGPT, for giving me regular updates on an album I thought we were building together). I saw the potential and loved the ideas, but in practical terms I saw limited real-world application combined with the risk that if you build a horizontal company, the next OpenAI and Anthropic launch might kill your company. I’m still bearish on the long-term durability of many incredible success stories like Midjourney and Lovable.

Then agentic AI arrived and pulled me further into the founder’s orbit. Suddenly, you could build practical applications, go vertical, VERY vertical, and apply LLMs to discrete, high-value problems while still being flexible enough to meet customers where they are. On top of that, you can turn startup execution efficiency up to 11. Still, finding an industry with a massive problem you know how to solve is rare.

Then an old friend pushed me over the edge. Matthew Kligerman and I have a history that sounds made up: an American climber (him) and a French tennis player (moi) met in Brazil, got into a car crash on the first meeting, then became basketball teammates. Besides being an incredible athlete, he is also an exceptional entrepreneur who built Escale.ai, one of the most successful Brazilian startups of the 2010s. He moved back to the U.S. and spent over a year figuring out his next adventure. When he launched Dirt.ai, I was surprised he chose to do it with a venture studio but once he introduced me to Stackpoint, it all made sense.

The Stackpoint partners (Adam, Chris, Matt) are hungry. They’ve launched 10 companies in proptech and construction tech in just a few years, and they’re highly founder-friendly because they’ve been in the seat themselves. I was impressed by the depth of diligence, the number of ideas they killed after diligence, and the reach they have in their verticals with customer access and proprietary datasets that typically take years to earn. As soon as they walked me through the construction pain points, I was hooked.

Construction + Agentic AI

Construction is the largest, most consequential industry in the world. It is also one of the hardest operating environments to get right.

Every commercial project is a coordination challenge across developers, architects, owners, PMs, supers, engineers, procurement teams, and subcontractors across multiple trades. Most people involved are highly skilled. The issue is not effort. The issue is orchestration.

Information is scattered across drawings, specs, submittals, schedules, emails, RFIs, meeting notes, and field updates. Teams are spread across companies with different incentives, different timelines, and different priorities. People are managing multiple jobs at once. In that reality, misalignment is not an exception, it is expected. And when it gets discovered late, it gets expensive.

That’s why rework represents a $175B+ annual loss in U.S. construction.

Beyond being a massive market, this is an efficiency play that combines compliance and risk detection.

Many SaaS companies have tried to tackle this, only to cap out around $5M ARR. At the time, the technology required people in the field to adopt a new platform, new training, and new logins. Worse, because project teams change every time, using a new tool meant retraining new partners repeatedly or getting value on only a subset of projects.

Agentic AI changes this. We can now build solutions that mesh with established workflows without forcing people to change habits. We can meet users where they are and adapt to how they work.

That creates the opportunity to reduce coordination drag, surface risk earlier, and improve decision quality before issues become field mistakes, schedule slips, and margin erosion. If executed correctly, we can materially reduce rework and increase decision precision across commercial projects.

Huge market, problems I love to solve, and the chance to build a truly efficient product and company. It felt like a perfect fit.

After weeks speaking with hundreds of industry experts and embedding the company with multiple design partners, I could not be more energized. I keep testing, building, learning, and building again. We now have our wedge product and are about to do our first customer onsite.

People who choose construction, the PM who chose this over finance, the engineer who chose this over software, take a pay cut to do what they love: build things that last. We’re building the platform that takes the friction away so they can do exactly that. And I want to find the people who feel the same hunger I do about making it happen. It’s time to build a stellar team.

If you’re a construction operator who lives in the details of submittals, RFIs, schedules, and procurement, I’d love to talk.
If you’re an engineer who wants to build real AI products in a massive vertical, reach out.
We’re building something ambitious, and we’re just getting started.

LET’S FUCKING GO.

Olivier Grinda

Entrepreneur, Tennis player, Gamer Know-it-all with a good heart.

I have launched and scaled 4 companies, raised over $60M in capital and had 2 successful exits in the last 15y. As COO of Clearco, I led a full operational pivot, deploying over $500M in capital, rebuilding product, risk, and data systems, and stabilizing the business. I am also an active angel investor in more than 70 companies.