Maive Thinks Your Roofer Is Leaving Money on the Table

The Macro: Home Services Is a $600 Billion Industry That Still Runs on Gut Feelings

Roofing alone is a $60 billion market in the U.S. The average roofing contractor runs a business doing $2-5 million in annual revenue, employs 10-30 people, and manages their estimates with some combination of measuring tape, satellite imagery from EagleView, and personal experience. The estimate is where the money is made or lost. Underbid and you eat the cost overrun. Overbid and you lose the job to the contractor down the street.

The existing tools for contractors are functional but not smart. CompanyCam is a photo documentation platform. Hover does 3D property models from smartphone photos. JobNimbus and AccuLynx are CRMs with project management features bolted on. EagleView provides aerial measurements. These tools help contractors organize their work, but none of them actually tell a contractor “your estimate is wrong and here’s why.”

The revenue leakage problem is real and measurable. Industry data suggests contractors leave 10-20% of potential revenue on the table through inaccurate estimates, missed line items, and inconsistent pricing. On a $15,000 roofing job, that’s $1,500 to $3,000 walking out the door. Multiply that across 500 jobs a year and you’re looking at a contractor losing $750,000 to $1.5 million annually. Nobody is seriously solving this with AI yet.

The Micro: Robotics Engineers on Your Roof

Maive is building an AI automation platform for home services contractors, starting with roofing estimates and quality control. Their website says “repair revenue leaks in your estimates,” which is a specific and compelling value proposition. The product analyzes estimates using AI and flags errors, missing items, and pricing inconsistencies before the contractor sends the bid to the homeowner.

The founding team is absurdly overqualified for roofing software, and I mean that as a compliment. David Tondreau, the CEO, was a manufacturing engineer at Rolls-Royce, then a senior manager at Dexterity AI (a robotics unicorn), holds a Master’s in Engineering from UC Berkeley, and has 15 patents in robotics and aerospace. Will Cray, the CTO, was a computer vision and ML tech lead at Dexterity AI, has a Master’s in ML from UT Austin, and started his career at Intel working on autonomous driving. These are people who’ve built computer vision systems that guide robotic arms in warehouses. Applying that expertise to roofing estimates is like hiring a Formula 1 engineer to tune your Honda Civic. The Honda is going to run very well.

They’re a two-person team in San Francisco, part of YC’s Winter 2025 batch. The narrow focus on roofing is smart. Home services is huge but fragmented. Roofing has specific, well-defined estimation workflows that are similar across contractors. Win roofing, prove the model, then expand to HVAC, solar, siding, and gutters.

The Verdict

I like Maive more than I expected to. The “AI for boring industries” playbook has produced some of the best venture returns of the past decade, and home services is about as boring and lucrative as it gets. The specific focus on revenue leakage is the right hook. You’re not asking a contractor to change how they work. You’re telling them you can find the money they’re already losing. That’s a much easier sell than “here’s a new platform to learn.”

The risk is adoption. Contractors are famously resistant to new software. The home services industry has a technology adoption curve that lags behind almost every other sector. Getting a roofing contractor who’s been bidding jobs from memory for 20 years to trust an AI estimate review tool is a sales challenge that has nothing to do with the quality of the technology. The team will need to show ROI in dollars, not features, and they’ll need to do it fast enough that contractors don’t churn during the learning curve.

There’s also the question of data. Roofing estimates vary by region, by material costs, by local labor rates. An AI that works for contractors in Dallas might give bad recommendations in Minneapolis where the roofing requirements are completely different. Building a model that accounts for geographic variation at a granular level takes time and a lot of training data.

In 30 days, I’d want to see contractors actively using the product on real bids, with before-and-after data on estimate accuracy. At 60 days, revenue recovery numbers matter. If a contractor can point to a specific job where Maive caught a $2,000 error, that’s a testimonial that sells the next hundred accounts. By 90 days, the expansion signal will be whether contractors are asking for Maive to cover more than just roofing. If the platform proves itself on roofs, every other trade is the same problem with different materials. The team has the technical depth to build this right. Now they need to sell it to people who’d rather be on a roof than behind a screen.