Why We Invested in Bobyard: AI Takeoffs for the Next Generation of Builders

Bobyard uses computer vision and AI to turn messy construction drawings into fast, accurate takeoffs and estimates—freeing contractors to bid on more work, with more confidence, and less overhead. Primary is proud to be the lead investor in Bobyard's seed round and are excited to continue to support founder Michael Ding and the team as they announce their $35 million Series A, led by 8VC.

The bottleneck between drawings and bids

Every construction project starts with a plan set. Before anyone can break ground, someone has to turn those drawings into quantities of material, labor, and equipment, then into a scope of work and a bid.

For mid-sized and large general contractors, that work falls to large teams of full-time estimators both on and offshore. For smaller firms, it often lands on owners and project managers who are already buried in work and hate the idea of counting items page by page with archaic software .

Takeoff work is slow, error-prone, and constrained by people, not demand. The construction industry is already short on estimators, and that gap is widening. Even as employment growth stalls, thousands of estimator roles need to be filled each year just to keep up. Contractors feel that constraint every time they want to bid a new job and realize their estimating team is tapped out.

Legacy takeoff tools help a bit, but they still expect humans to “teach” the software where everything is on the drawing: tracing rooms, placing shapes, counting symbols by hand. They facilitate the workflow; they do not do the work.

Bobyard flips that around.

Why now: computer vision that actually does the takeoff

Bobyard’s core product is an AI-native takeoff engine that has helped contractors increase the number of bids they do over 5X. With Bobyard, contractors upload a PDF, select the legend, and run analysis. Under the hood, dozens of computer vision models—trained specifically on construction drawings—spin up in parallel to:

• Detect symbols and count them
• Measure linear footage
• Segment and measure areas
• Parse text and notes

Instead of forcing estimators to drag polygons over every planter bed or conference room, Bobyard’s proprietary models recognize each distinct region, size them, and calculate quantities automatically. First-pass accuracy is already in the mid– to high–ninety percent range for the initial trades, with more than ninety percent of the work fully automated.

On top of that, Bobyard handles scope-of-work generation and estimation: material and labor costs, templates, markups, and exports that plug into existing workflows. It is a complete workflow from drawing to bid.

The impact for customers is straightforward: takeoffs that used to take dozens of hours can be turned around in a fraction of the time, with better consistency and fewer misses. That lets each estimator support more bids, say yes to more opportunities, and grow revenue without adding headcount.

Meeting Michael Ding: from math olympiad to front office painkiller

We met Bobyard founder and CEO Michael Ding through Pear VC’s accelerator.Mar Hershenson knew about our thesis and deep connectivity across the built world and flagged him as one of the strongest founders to come through their program.

Michael grew up in the Bay Area, started coding at six, competed in math olympiads, and graduated valedictorian from one of the most competitive high schools in the region. At Stanford, he focused on computer science and AI, but his real education came from the field: in three months, he spoke with about 250 general contractors to understand how they actually estimate work.

He did not do this from behind a desk. He cold emailed. He called. He stood at Home Depot at 7 AM and talked to contractors one by one. Aside from proving a level of obsession we look for in Founders, those conversations also pointed to the same pain time and time again: takeoffs and estimates were the bottleneck.

Only after validating demand did he start building. In roughly a month, with occasional help from a roommate, Michael reproduced and improved on the capabilities of well-funded AI takeoff competitors. From there, he began training models trade by trade—starting with walls, paint, flooring, windows, doors, and hinges, and pushing into more complex categories over time.

What stood out to us:

• He is extremely structured in how he learns and decides what to build.
• He combines deep technical ability with real hustle on sales and customer discovery.He moves fast, but is careful about accuracy and reliability—exactly what you want in software that controls millions of dollars of bids.

A big, underestimated market hiding in the “front office”

On the surface, takeoffs and estimating might look like back-office work, but in reality is the key unlock to revenue.

In the U.S. alone, construction companies spend billions each year on estimator labor. Many large GCs also outsource a significant portion of estimation to lower-cost markets. Smaller contractors sacrifice nights and weekends to get bids out the door. That is all “hidden spend” on the same underlying task: turning drawings into numbers you can sign.

Long term, we believe Bobyard can become the operating system that connects drawings, takeoffs, bids, and, eventually, more of the construction value chain. The team is quickly expanding into all major construction trades, and takeoff is the foundational data layer that unlocks every other piece of preconstruction.

We are excited to support Michael from this early stage as he and the team build toward that vision—and as Bobyard becomes the default way contractors turn plans into projects.