Five corporate innovation leaders from Vinci, Leonard, and Kajima sat down with Bricks & Bytes in Paris. The throughline was the same in every conversation: you don’t sell construction tech to a giant contractor, you sell it to one business unit, on one project, by solving a pain they already feel. Here’s how the people who actually sign the checks think, and what gets a founder a fast no.
Bricks & Bytes ran a three-hour livestream from Leonard’s demo day in Paris, the foresight and innovation platform of the Vinci group. The guests included Julien Villalongue, Managing Director of Leonard; Kevin Cardona, Deputy Director at Leonard; Yogesh Patel, Quality, Improvement and Innovations Director at Vinci Construction UK; Guillaume Malochet, Global Director of Strategy, Innovation and Transformation at Vinci Construction; and Yuji Doi, Innovation Manager at Kajima. Different roles, different countries, one shared headache: getting good technology adopted inside enormous, slow-moving organizations.
If you build or sell ConTech, this is the buyer’s brain laid bare. If you run innovation inside a contractor, it’s a look at how your peers climb the same wall. The lesson lands the same either way: the org chart you pitch is almost never the org chart that buys.
Stop trying to land “Vinci.” There is no single buyer to land. The real game is winning one project manager, proving value on day one, then letting that win travel between business units on its own.
There’s no single front door
Why one Vinci contract doesn’t exist
Villalongue was blunt about the structure. Vinci isn’t a monolith, it’s a federation of roughly 4,000 companies, what he calls little countries, or an archipelago. People in the field own their projects and their innovation decisions. So the founder fantasy of signing one deal at the top and rolling out to 300,000 people is exactly backwards. You convince different business unit managers, in different countries, one at a time.
Cardona made the same point from procurement. Leonard does not centralize purchasing, because purchasing happens at the business unit level. There is no contract between one startup and the big Vinci, he said, because that does not make sense in a fragmented sector. His team’s job is matchmaking: connect an outside founder with the internal operator who feels the pain, then get out of the way.
We don’t sell to Vinci. That doesn’t make any sense. We help two entrepreneurs talk to each other.Kevin Cardona, Deputy Director, Leonard
Problem-pull beats tech-push
Lead with the pain, not the product
Cardona kept returning to one phrase: problem-pull, not tech-push. The fastest way to get cut is what he calls the false good idea, where the team loves a founder and dreams up a use case that does not answer a real pain. To avoid it, Leonard’s Design Lab runs workshops that open with “talk me through your process,” never “let’s talk about AI.” Technology enters only once a genuine bottleneck shows up.
Villalongue named the two mistakes he sees most when founders try to sell into construction. First, treating Vinci as one buyer. Second, a product mindset instead of a deployment mindset, pushing a clever product rather than fitting a messy operational reality. On a live site, a team already has a stack of problems. If your tool becomes another one, they quietly forget it.
- Treating the giant as one buyer. There is no global yes. Plan to win business units one by one.
- Product mindset over deployment mindset. Selling features beats nothing. Fitting the on-site workflow beats everything.
- The false good idea. A use case you invented for a pain nobody has gets a polite, fast no.
Prove it on one project, on day one
The death valley nobody warns founders about
Here’s the trap that catches polished pitches. Villalongue explained that decisions get made project by project, so you have to prove value for a single project from day one. If your ROI only works spread across a pool of projects over ten years, it falls apart, because the person deciding is looking at one job, one budget, one schedule. The math has to close on its own, right away.
Cardona calls the moment after selection the death valley. Passing Leonard’s filters is not the real exam. The real exam is whether someone will actually pay for the thing, because a payment proves an urgent need rather than a nice-to-have. That bar is brutal, which is why the survival rate is worth noting: around 60% of the startups Leonard has accelerated were still alive five years later, well above the usual mortality curve.
The real test is whether someone will pay for that somewhere. If they pay, it answers a real need.Kevin Cardona, Deputy Director, Leonard
Get the construction tech briefing founders and execs actually read
One sharp take a week on who’s building, buying, and funding ConTech, straight to your inbox.
Join 3000+ ReadersThe pilot is easy. Scaling is the wall.
How giants turn one win into a standard
Patel was direct about where innovation dies. There are plenty of pilots, he said, and everyone plans for failure. Far fewer ask what happens if a trial works. Scaling that win from one site to every site, and baking it into processes and standards so it becomes business as usual, is the genuinely hard part. Vinci Construction UK leans on innovation champions in each business unit, plus internal awards that have run since 2006 and now pull more than 250 entries. None of it appeared overnight.
The Cobot is a tidy example. It started as a robotic cutter for gypsum board and met skepticism, because the tool worked but the process around it had not changed. Once teams adapted, its use widened to facade and concrete board cutting, and Vinci UK is buying a second. Malochet, a sociologist by training, frames scaling as change management more than engineering. People do not love being the second to adopt something, so much of his job is convincing them that what worked next door will work for them.
Cardona described that last step as the real engine. When a team hesitates, Leonard points them to a colleague on another continent who solved the same problem with the same startup. A short peer-to-peer call, where that colleague explains the ROI, beats any central mandate. Operators convince operators.
AI and robotics: real, but earned
The shift from software to hardware
Villalongue has watched the technology change since Leonard started working on AI in 2017. The bigger story now is the shift from software to industrial hardware, with robotics sitting roughly where AI sat a decade ago, early but climbing fast. Cardona pushed further, half-joking about the end of SaaS: once AI removes the barrier to building software, value moves to whoever understands the use case best, and a lot of the interesting work picks up a hardware and robotics dimension that needs heavier, CAPEX-style financing.
Malochet gave the honest version of corporate AI adoption. He was reluctant at first, until the company’s CFO told him to start using it. His rule now is to find a real use case, try it, and drop it if it does not deliver. The automated quarry dumper Vinci is testing in Quebec came from a small company, not Caterpillar, and the drivers it displaces get redeployed to run the quarry remotely rather than shown the door. From Tokyo, Doi added that Kajima is leaning into physical AI so robots can read a changing site and adjust instead of stalling, alongside deep-tech sensing for its tunnel and renovation work.
That’s not AI for the sake of AI. You find a use case, try it, and if it works, fine.Guillaume Malochet, Global Director, Vinci Construction
What actually gets deployed
Problem-pull versus tech-push, side by side
Across five very different leaders, the contrast held. Founders who get in start from an operational pain and prove it cheaply. The ones who stall start from their own product and hope the industry reshapes itself to fit. Here’s the split, drawn from what they said on stage.
| Decision point | Gets deployed (problem-pull) | Gets a fast no (tech-push) |
|---|---|---|
| Who you target | A business unit manager on a live project | Central HQ, hoping for a group-wide rollout |
| Where you start | A named pain the team already feels | A slick demo of what your product can do |
| How you prove value | ROI that closes on one project, day one | Value pooled across many projects over ten years |
| Your mindset | Deployment: fit the workflow on site | Product: ship features, expect adoption |
| How it scales | Peer operators vouch across business units | A top-down mandate nobody asked for |
| Your AI angle | Tied to a specific, testable use case | “We have AI,” with no problem attached |
Leonard is the foresight and innovation platform of the Vinci group, launched in 2017. It runs incubation and acceleration programs for both outside startups and Vinci employees, and connects them with the group’s operating businesses. It does not buy on the group’s behalf. Its role is to match the right solution to the right operational team. You can read about its programs on the Leonard site.
Because, as its leaders described it, Vinci operates as a federation of roughly 4,000 semi-independent business units, and purchasing happens at that local level rather than centrally. There is no single switch to flip. A founder typically has to win individual business unit managers, often in different countries, and let proof travel between them.
It means starting from a real operational pain that a team already feels, then finding technology to solve it, rather than starting with a product and hunting for somewhere to apply it. Cardona warned against the “false good idea,” an invented use case that sounds great but does not map to a real need. The fix is to ask operators to walk through their process before mentioning any tool.
The pilot is the easy part. Scaling means embedding the tool into standard processes so it becomes business as usual, which is mostly change management. Vinci Construction UK uses innovation champions in each business unit and long-running internal awards to spread wins, and leans heavily on peer-to-peer proof, where an operator who solved the same problem elsewhere convinces the next team.
Yes, though selectively. Vinci is testing an automated quarry dumper from a small supplier in Quebec, and Vinci UK runs a robotic cutting Cobot it is now expanding. Kajima is deploying physical AI so robots can adapt to changing site conditions. The common thread is that each deployment is tied to a specific pain, not adopted for novelty. You can read more on how robotics is entering construction in our robotics report.
Bringing a product mindset instead of a deployment mindset. Villalongue described founders who want to sell a clever product without understanding the constraints on a live site. If a tool adds friction to a team already juggling cost, schedule, and quality, it gets dropped. Proving value on one project, immediately, beats any pitch about long-term potential.
Leonard, Vinci’s foresight and innovation platform: leonard.vinci.com
Vinci research and innovation overview: vinci.com
Kajima Corporation (global): kajima.co.jp
Related Articles
The construction tech go-to-market guide
How one startup is bringing robotics into construction
