Data Integration, Use Cases and AI’s Next Frontier: Insights from Trimble’s AI Director

Trimble has done a great job of creating hardware to capture construction data, but now the challenge is making meaning out of it.

Determined to not be buried under the raw data, Trimble is one of many software solutions adopting and expanding its AI capabilities to make use of the copious amounts available. We had the opportunity to hear from Karoliina Torttila, director of AI at Trimble, who leads the company's AI research and development, and understands firsthand where AI software integration is helping contractors make the most of their data.

1331456353_Events_AI ML Online Forum_April Thumbnail.headshot image

Below is a recap of some of the valuable insights Torttila shared during her presentation:

Freedom in design leads to efficiency in design deliverables.

Torttila's team focuses a great deal on Trimble's design software and has noticed, "one of the most exciting things for us on the design side is how we are seeing the introduction of AI to different workflows as an enabler of freedom in design."  As design workflows often end up stuck on tables, AI is a valuable tool for taking away headaches in designing. We are starting to see design being approached by all kinds of data modalities: basic sketching, natural language (text), 2d images (single image reconstruction), going from floor plans to 3d designs and so forth.

Solving issues in one area allows for solutions in another as the same tech can apply across different use spaces.

The use cases for AI are diverse. As solutions are created to address problems in one space, it is found that those tools work for a lot of use cases. Torttila highlighted processing field data as an interesting area because it doesn't matter if the use case is retrofit, progress monitoring, asset maintenance, or end-of-life inventory, ultimately a lot of processing needs are the same technology broadly applicable across scenes and scenarios. She pointed to instances at Trimble where she sees the cross-application of solutions such as road asset inventory (an area where the company has put out a lot of deep learning models), building facades and column detection.

AI best use cases with repetitive work ripe for automation

AI applications interesting from an efficiency perspective are the ability to create design deliverables in an automated manner. Torttila noted an example they see is when an individual has spent a lot of their time creating 3D designs then they don't want to spend time creating drawings that might be needed on the regulatory side or detailing on the structural side. Repetitive work is ideal for AI automation. Another example is detailing, a laborious effort, making it a great use for AI. In these scenarios, AI can take past project geometry to rinse and repeat in new areas where it makes sense.

Balancing full automation vs using AI as an assistant

Right now, not all these processes are fully automated or need to be. In the long term, Trimble aims to put out tech providing as much automation as possible. However, it is important to note that semi-automated workflows are useful too. As Trimble releases and integrates a lot of these AI solutions into its portfolio, it is relying on users and customers to identify how much it can realistically and pragmatically automate at once and where it should introduce these verification points to deal with the unpredictability of AI processing.

Visual models and analysis next frontier

When it comes to construction management, one typically thinks of AI capabilities in terms of text-based analysis, but visual-based analysis is also a use case. Trimble is looking at the introduction of text to visual models. So far, the company has released products like SketchUp Diffusion to create efficient ways to produce creative renderings. Currently, SketchUp Diffusion is just a plus-up offering rather than creating value every day, however, there's potential for this type of technology to grow in use and value.

Insights from BuiltWorlds Research on AI/ML Solutions

AI/ML technology solutions are a growing part of the AEC tech toolkit. According to the BuiltWorlds 2023 AI/ML Baseline Survey, 58% of companies in our sample are utilizing software that offers AI/ML features aimed at enhancing core business functions. An additional 23% of respondents had plans to incorporate that technology. Out of those, only 13% of respondents use these solutions for every project, and an additional 6% implement them on most or a few projects. The largest segment of respondents, 38%, are implementing AI/ML solutions on a few projects. These results indicate moderate and growing adoption of AI/ML tech solutions in the AEC industry.