Why We Invested in Primepoint: Rebuilding Construction’s Operating System with AI

Construction is one of the most complex coordination problems in the world. And yet, much of how projects get built still relies on day-of human judgement, supplemented with a combination of PDFs, email threads, and “phone a friend” decision.

Every project — from a tenant improvement to a billion-dollar data center — is driven by drawings, specifications, submittals, RFIs, and schedules that communicate what to build, when, and where. These documents are deeply interconnected, constantly changing, and often interpreted differently by each stakeholder involved.

The result is predictable: miscommunication, delays, rework, and cost overruns.

At GS Futures, we look for companies that don’t just improve workflows at the margin, but rethink how work should happen entirely. Primepoint is one of those companies.

The problem isn’t software; it’s understanding the details

The construction industry already has thoughtful software. Platforms like Procore and Autodesk Construction Cloud have become systems of record for projects. They centralize documents, track workflows, and improve visibility. But they don’t fundamentally solve the hardest problem: understanding what’s actually in the documents.

A single detail in a drawing might require cross-referencing multiple sheets, specifications, schedules, and submittals. Even experienced project engineers spend hours navigating this information and may still miss things.

Primepoint makes life in the jobsite trailer better

Instead of manually reviewing drawings, writing RFIs, or checking submittals, teams can rely on Primepoint to:

• Automatically connect information across drawings, specs, and documents
• Flag constructability issues before they hit the field
• Draft RFIs based on incomplete or conflicting information
• Review submittals against project requirements — and connect these to the source of record (i.e. Autodesk Construction Cloud)
• Answer project-level questions through natural language

To do this accurately, effectively, and quickly, Primepoint is building a construction knowledge graph — a system that understands how documents relate to each other at a granular level, and can trace every answer back to its source.

Primepoint’s approach — combining computer vision with structured understanding of drawings — is what enables that level of reliability. This is an incredibly hard challenge, which requires contextual understanding beyond what a mass-market VLM and other commercial AI tools can deliver.

Why this is a hard (and important) problem

One of the reasons we were drawn to Primepoint is that they intentionally chose a problem that is difficult.

Construction drawings are not standardized. They vary by discipline, by firm, by project type, and by geography. They contain dense symbols, annotations, and references that require domain expertise to interpret.

Get Aaron Toppston’s stories in your inbox

Join Medium for free to get updates from this writer.

Subscribe

Remember me for faster sign in

General-purpose language models aren’t designed to interpret visual engineering data. OCR-based tools extract text but miss meaning. And many existing solutions stop at search as they don’t actually understand relationships between documents.

Primepoint is taking a different path: building proprietary computer vision models that interpret drawings at a deep level and connect them into a unified system. The implications of this ground-up foundational strategy go far beyond a single workflow.

From point solution to platform

Today, Primepoint is solving four high-value (and well-known) workflows:

• RFI generation
• Drawing navigation
• Constructability analysis
• Submittal review and orchestration

But the long-term vision is much larger.

The same underlying technology can power a full AI-first project management platform — one where repetitive administrative work is automated, and institutional knowledge compounds over time.

In construction, knowledge is often trapped within individual teams or lost between projects. We have learned this first hand; the first job of most PEs is to review drawings and learn through the motions that experience (not formal education) provide. Primepoint’s knowledge graph creates the foundation for that knowledge to scale across an entire organization. This organizational approach is particularly beneficial for new talent entering the industry as the learning curve is flattened and time in the field increases.

Why we believe Primepoint can win

There are a few reasons we’re particularly excited about this investment:

1. Exceptional technical team
This is one of the strongest AI teams we’ve met in construction. Lubomir (co-founder) helped build Facebook AI Research and has an incredibly deep background in applied AI, while Hamid brings product experience from Trello and Atlassian. Kamran joined as an experienced construction professional — creating an rare combination of deep research + product execution + industry knowledge.

2. A real technical moat
Primepoint’s advantage is it’s the underlying ability to interpret and connect construction documents. This is fundamentally hard and not easily replicated with off-the-shelf models. The depth of their proprietary model allows client on-boarding quickly and mid-way though a project, which is equally important to seeing results in the field. This technical moat equally enables Primepoint to serve multiple stakeholders across the project value chain in the long run.

3. Early product validation
Even at an early stage, customers are seeing meaningful value. In pilot projects, Primepoint has identified real design issues and improved constructability workflows in ways that competitors have struggled to match. In construction, where skepticism toward new tools is high, that level of early validation matters. Ultimately, it is this effectiveness that helps cut through the noise of multiple solutions that sound the same, but are fundamentally different.

Looking ahead

One of the themes we continue to see across our investments is the shift from software as a system of record to software as a system of execution. Primepoint is a clear example of that shift. Instead of simply storing information, the platform actively interprets it, connects it, and takes action on behalf of the user. We believe this ethos is critical to making construction a better industry — for project outcomes and the day-to-day experience of those building.

We’re excited to partner with Primepoint on this journey.

‍