Plangrep, Others Offer Construction AI-As-A-Service (CAIaaS)

CAIaaS Can Be Embedded in Construction Software or Individual Contractor Workflows

On The Explainer blog, we have focused on the fact that large language models (LLMs) are not the begin-all and end-all of construction AI (artificial intelligence). More deterministic approaches to prediction, orchestration, scheduling and machine vision have enormous potential and already are making great strides.

While construction software vendors have opened their applications to consumer-grade AI models, more focused, construction-specific LLMs are emerging that could notch these out some of that spend plus some giveback, driving more value, while using fewer resources.

Zetane is an early entrant to this market—I’d encountered them back in 2022 as their initial takeoff product spun out of a project with Pomerleau. In April of 2026, Zetane landed $1.8 million in a seed round after bootstrapping since its inception.

I found myself thinking of Zetane a while ago when we got a briefing from Minneapolis-based Plangrep. According to  and Cofounder Tom Juntunen and CEO/Co-Founder Anastasiia Juntunen, their offering turns plans, specs, addenda, and Office files, and other documents are ingested and ported into JSON-driven:

• Requests for information (RFIs)

• Scoping work

• Revision change tracking

• Bid leveling

• Constructability reviews

• Visual mockups

• Highly designed Excel, Word and PowerPoint files

The company has two essential offerings—a streamlined AI workspace for architecture, construction and engineering (AEC) and an application programming interface (API) that supports construction software companies and larger AEC customers with their own building information modeling (BIM), virtual design and construction (VDC) or devops teams building out AI features.

Modern construction software should be built on RESTful APIs (more here) which contain well-structured descriptions about the information they expose to other software products. These RESTful APIs support process orchestration, analytics, and now large language models that can not just expose data, but use the application to take action based on that data. Plangrep layers on its own construction-informed metadata into a JSON file (which is coin of the realm for agentic AI), making AI inference and action quicker, more focused and more efficient with AI tokens.

What is Plangrep?

Plangrep is not so much a construction software application as it is a construction artificial intelligence as-a-service (CAIaaS) platform and API that transforms unstructured construction drawings into structured, machine-readable data for AI systems, software applications, and human users. Tom Juntunen compared it to Firecrawl, which prepares web content for AI ingestion, but for construction data.

Plangrep is agnostic, and can add value to ChatGPT, Claude, Gemini and open-source models.

"The problem is that AI doesn't have the spatial awareness needed to work with large volumes of construction drawings,” Anastasiia Juntunen said in our recent briefing. “Our technology helps AI, and as a result, the end user’s AI can understand these documents the way a human would."

While AI applications using live transaction data must reference the data source in real time, a static plan set can be converted by Plangrep into an AI-optimized JSON reference file.

"We structure the data from construction drawings and store it in a database, so AI doesn't have to keep reading massive PDF files,” Tom Juntunen said. “It can retrieve exactly what it needs."

By saving work for (and token usage by) AI, Plangrep significantly speeds up data extraction.

"We can produce a revision log for a 200-page drawing set in about 30 seconds,” Tom Juntunen said.  “That's unheard of."

Plangrep Market

Right now, the founders say they are selling primarily to construction software companies, including product development and engineering teams at construction software-as-a-service (SaaS) startups and established construction software vendors.

These construction software companies can use Plangrep to speed to market offerings like AI assistants, project intelligence tools, asset management systems and procurement applications.

While the API-centric nature of the product makes Plangrep a natural technology for construction software vendors, construction contractors are buying in as well. Plangrep is being leveraged currently by commercial contractors, architects, estimators and preconstruction teams.

Tom Juntunen described his experience doing overlays with a legacy product, describing it as slow and also problematic because between revision sets, architects would make changes requiring the entire sheet to be reindexed instead of just the portions that were changed. This is one reason creating a central JSON index independent of the PDF, that enforces consistency, enhances speed to such an extent.

Project and asset owner-operators can also use Plangrep to tame large numbers of as-built and as-maintained plans in fields like utilities, power generation and energy.

How Much Does Plangrep Cost?

Plangrep is on a freemium model for its AI workspace for AEC, and a free plan can be topped off for as little as $10 to $50. Other models include monthly subscriptions and pricing based on consumption of the AI.

The founders say top-off payments can be as little as $20. Subscription plans run $200 per month to start, which will handle more than 1,300 drawing pages or more than 13,300 document pages. Pricing becomes more negotiable with volume.

Most customers purchase directly on the web site, in a developer-driven motion rather than an enterprise software sales cycle with no pilot required.

BOTTOM LINE: The velocity of AI in preconstruction and construction operations depends on how well software vendors and internal dev ops teams can mitigate risk. How can they present clear wins with minimal cost and predictable risk? Trying to re-invent a construction-informed approach to AI yourself is probably not the way to do any of those things. So CAIaaS may be a category to watch.