ChatGPT for construction offers faster takeoffs, sharper estimates, less waste


Sparkel is an Oslo-based startup that builds software for helping construction professionals interpret and understand construction drawings using artificial intelligence.

The Challenge: Contractors waste time and money due to poor insights from design documents

More than a hundred design documents can be generated for a large construction project, with new versions created frequently. Many contractors risk misinterpreting the designs because files arrive as static PDFs and spreadsheets and/or miss critical construction data. The consequences include:

  • Under- or over-estimating the project’s cost and schedule
  • Loss of bid or loss of money
  • Material waste due to inaccurate quantities

Construction professionals such as estimators, project managers, and site managers need a solution that links design documents to reality. They need to pull updated data from building information models (BIM) and PDFs to spot design problems and opportunities, reliably estimate quantities, respond to design changes, and communicate with their stakeholders.

APS Solution: Sparkel layers on ChatGPT and other AI to extract construction document insights

The solution is Sparkel’s cloud-based software, which deploys the Model Derivative API to extract and translate 2D and 3D designs and the Viewer to display models in a web browser. Sparkel layers on AI to analyze the drawings and ChatGPT to search for results. With Sparkel, construction professionals can:

  • Ask questions about BIM models and PDFs using natural language, eliminating the need for technical training
  • Estimate quantities and generate takeoffs directly from the information in the design documents, reducing errors
  • Identify problems with the design and increase sustainability by reducing waste and analyzing the carbon footprint
  • Communicate with suppliers and subcontractors with the same data, reducing lag time and errors

Ask the models questions such as, “How much precast concrete is on the 2nd floor?” and Sparkel’s AI engine will present tables and information directly linked to the drawings.  When design updates arrive, Sparkel analyzes and communicates the changes. This seamless link to construction enables more accurate models that can be used as the basis for a digital twin at handover.

Sparkel’s ultimate goal is to enable circularity. With the digital twin holding precise information about the building and its materials, that data can be used to inform material reuse during renovations or demolition.

AI-powered collaboration delivers insights and produces results

  • Accurate cost estimates help win the right projects and reduce risk
  • Sparkel customers estimate takeoffs happen 5x faster than manual processes and a savings of more than $40,000 per project by identifying problems in design
  • Sparkel supports multiple large models, saving customers time and money maintaining desktop installations and licenses
  • Digital twin at handover adds value, increases contractors’ competitive edge

“No-waste construction is possible. But to get there, the industry needs to better translate data from the design phase into the construction phase. The data needs to work for the realities of construction processes and the job site. This is how we improve cost estimates and margins, reduce waste and error, and move closer to sustainability.” - Magnus Nilsen, CEO of Sparkel

APIs and Services

Model Derivative API
Model Derivative API

Autodesk Products


Related Stories

Ready to Get Started?

Try Autodesk Platform Services for free by setting up an account and free trial. Explore the resources to learn about the solutions you can build using design and engineering data.


Try APS for free