Navitas Capital - Why We Invested in Primepoint

(Navitas Capital) - The best AI talent in the world is typically found building foundation models, self-driving cars, and image generators. Historically, the $13 trillion construction industry hasn’t been a magnet for Silicon Valley AI researchers. That’s starting to change as the complexity of construction’s data challenges draws a new wave of technical founders.

Andreessen Horowitz recently published a blog post, arguing that the architecture, engineering, and construction industry represents one of the largest and most compelling opportunities for AI-native software. We agree and have been investing with conviction in the segment for over a decade, with the most AI exits of any VC investor in construction.

Most startups in the AEC industry today offer products with some AI capabilities. But the Primepoint team is unique in their true AI research chops that support a novel, differentiated offering for the market. We have seen products that demonstrate genuine technical depth, like our portfolio company Firmus (acquired by Nemetschek in 2025), command premium pricing and rapid market adoption.

So when we met Lubomir Bourdev and Hamid Palo, the co-founders of Primepoint, we paid attention.

World Class AI Talent

Lubomir and Hamid bring a rare combination of backgrounds to construction technology. Lubomir is a founding member of Facebook AI Research who built the original object recognition system deployed across every photo and video on Facebook and Instagram. He’s authored ~100 patents and accumulated ~100,000 citations. After Facebook, he co-founded WaveOne, the real-life “Pied Piper,” which Apple acquired in 2023 and now powers the iPhone’s video infrastructure. He could have done anything next. He chose construction.

Why? Because after speaking with contractors, he recognized that the industry’s most painful bottleneck — the manual, error-prone interpretation of complex construction drawings — was a problem perfectly suited to his expertise in computer vision and deep learning.

Hamid brings a different but equally impressive skill set. He joined Trello as employee #5, helped scale it into a globally adopted platform, and led the product through its acquisition by Atlassian. At Atlassian, he led Trello product teams before moving into a Special Projects role reporting directly to the Atlassian President. Hamid knows how to turn powerful technology into collaborative software that teams adopt organically, which is exactly the challenge Primepoint faces in bringing AI to construction project teams.

Rounding out the leadership team is Kamran Azarbal, VP of Strategy, who brings over a decade of commercial construction experience from Webcor, where he worked on multi-billion-dollar commercial projects. Kam has lived the pain that Primepoint is solving. He has spent years navigating the exact types of complex drawing sets and document workflows that the platform now automates, and his domain expertise sharpens every product decision the team makes.

Everything Is Connected

Here’s the thing about construction drawings that most people don’t appreciate: a single door on a floor plan can reference a legend, a detailed inset drawing, a door schedule with dimensions, a product specification for materials, and a submittal package for the hinge and jamb composition. That information is scattered across hundreds of pages of PDFs. Finding and reconciling it takes experienced project managers hours of painstaking manual work.

Large-language models can’t reliably do this. They lack the domain-specific training to parse linework, interpret annotations, and trace cross-references across document types. This is why so many “AI for construction” products fall short in practice. They can extract text from a PDF, but they can’t actually understand a drawing.

Primepoint can. The platform builds a proprietary knowledge graph that connects every drawing element to its corresponding schedule, specification, and project documents. It doesn’t just answer questions about drawings. It understands the relationships between them. The result transforms static PDFs into a seamless, interactive, visually connected experience where project teams can navigate between a floor plan annotation and its full specification context in a single click.

When we saw the demo, it was immediately clear this was different from anything else in the market. This is way more than a static report of design issues. Primepoint dynamically hyperlinks drawing elements to relevant information from other project documents (e.g., specifications, schedules, etc.) and can interpret key details that are not clearly spelled out in text. As early investors in PlanGrid, we’re familiar with the power of putting digital drawings in the hands of construction teams. But Primepoint goes much deeper in a way that feels like magic but is grounded in serious AI research.

From Reading Drawings to Running Projects

What makes Primepoint’s approach particularly compelling is that the knowledge graph isn’t just a search tool. It powers real workflow automation. The platform automates constructability reviews to catch design errors before they become costly field issues, generates and reviews RFIs grounded in actual drawing data, and streamlines submittal analysis against project specifications. These are workflows that every general contractor performs on every project, and they’re almost entirely manual today. In contrast to other technical approaches, Primepoint’s comprehensive understanding of project documents unlocks a broader set of discrete, but interconnected, automated workflows.

The a16z piece quantifies the cost of this status quo: construction professionals spend over 14 hours per week on non-productive activities, and rework driven by design errors costs the U.S. industry $177 billion annually. Primepoint attacks that waste at the source by giving teams a tool that actually reads and understands the documents on which their projects are built.

Early customers are feeling it. An ENR Top 50 general contractor independently produced a case study highlighting Primepoint’s impact, cutting drawing navigation time by over 50% while proactively surfacing design risks that would have led to rework.

A New Age, a Decade in the Making

At Navitas, we’ve spent over a decade investing at the intersection of construction and artificial intelligence. Our portfolio includes PlanGrid (acquired by Autodesk), Matterport (acquired by CoStar), Firmus (acquired by Nemetschek), Document Crunch (acquired by Trimble), and OpenSpace, each representing a different chapter in how technology can transform the built environment.

Primepoint is the next chapter. Where PlanGrid digitized drawings, and Firmus proved AI could assess them, Primepoint builds a genuine understanding of construction documents and uses that understanding to automate the workflows that consume project teams’ days. It’s exactly the kind of Vertical AI company we look for: world-class technical talent applying proprietary AI to a massive industry with clear, measurable ROI for customers.

We are excited to lead a round that also includes angel investor Dr. Yann LeCun, widely recognized as one of the founding fathers of modern deep learning, alongside co-investors Penny Jar Capital, NextView Ventures, GS Futures, and Aglaé Ventures.

We couldn’t be more excited to partner with Lubomir, Hamid, Kam, and the Primepoint team. They bring exactly the kind of deep AI research background this problem demands, paired with the construction expertise to ensure the technology lands where it matters.

Construction is ready for its AI moment. With Primepoint, we believe it’s arrived.

If you’re a general contractor or project owner looking to transform how your team works with construction drawings, visit primepoint.ai to learn more.

‍