In the world of building digitization, we’re standing at the intersection of legacy documentation and cutting-edge AI. Whether it’s architectural plans, HVAC schematics, commissioning handover documents, or electrical single-line diagrams, every building tells a story. The challenge? These stories are often buried in disparate, static documents. That’s where AI-powered subgraph generation presents an opportunity to increase onboarding speed and scale by orders of magnitude.

What is AI subgraph generation?

At its core, subgraph generation is the process of extracting structured knowledge from unstructured or semi-structured documents. When speaking in the context of digital twins, think of it as building a “brain” – something that understands the relationships between systems, components, and spaces in a building.

Using AI, we can now intake virtually any document type that describes a building and convert it into a structured format. This includes a variety of documents, from test and balance reports to renovation records from buildings constructed in the 1920s. If the document is in digital form, it can be processed.

The following figures show the workflow of populating a knowledge graph from documents:

Fig. 1: Populating the knowledge graph using AI.
Fig. 2: Input – A sample architectural drawing of an HVAC system.
Fig. 3: Output – A completed subgraph (nodes and relationships) generated in Willow.

Fast onboarding matters now more than ever

It’s no secret that the traditional approach to digitizing buildings is labor-intensive. It’s a tedious process that involves manually sifting through documents, highlighting key data, and transferring it into spreadsheets or databases. This process can take weeks, especially for complex facilities like hospitals with multiple HVAC zones, exhaust fans, and air exchange requirements.

AI changes the game. Willow can reduce weeks and weeks of manual work to a matter of hours—or even minutes. Our proprietary framework parses documents, extracts relevant entities, and maps them into a knowledge graph. This enables faster onboarding, better insights, and more accurate digital twins.

Robust documentation, better performance

Consider the commissioning phase of a building, when documentation is handed over and systems are verified. By applying AI at this stage, Willow can automate the creation of subgraphs that represent the building’s operational logic. This not only accelerates deployment but also ensures that the digital twin reflects the building’s true state.

Even buildings with incomplete documentation can benefit from it. Generative AI can work with partial data sets, infer missing relationships, and continuously update the graph as new information becomes available.

Complexity and scale

The complexity of digitization scales with the building’s size and typology. A warehouse might have a simple HVAC setup, while a hospital could involve dozens of VAV zones and specialized air handling requirements. AI helps manage this complexity by automating the extraction and structuring of data across disciplines and document types.

 Summary and key takeaways

  • As buildings become smarter and more complex, the need for accurate, scalable documentation grows but traditional methods can’t keep up. Manual data entry is slow, error-prone, and often outdated before it’s even complete.
  • AI-powered subgraph generation changes the equation. Willow transforms static documentation like plans, schematics, and reports into structured, dynamic knowledge graphs. This creates a living, searchable foundation for your digital twin, no matter the age or complexity of your building.
  • Documentation becomes an asset, not a bottleneck. From commissioning to retrofits, Willow automates the process of identifying systems, mapping relationships, and surfacing insights faster than ever before.
  • Even with incomplete or fragmented records, generative AI can fill in the gaps. It continuously learns, adapts, and improves, helping you keep a digital twin up-to-date.
  • Whether you’re managing a warehouse or a hospital, Willow scales to match the complexity of your environment. AI removes the manual overhead so your team can focus on the aspects that matter like performance, sustainability, and long-term value.

As we continue to improve how we leverage AI and expand our ontology, the goal is clear: make building digitization faster, smarter, and more scalable. Can leveraging AI for subgraph generation help you unlock the full potential of your building?