What is agentic AI and is it powerful enough to change construction’s future?

Premium Content
Listen to this article

Could artificial intelligence (AI) be about to make another leap – one that could profoundly change workflows in construction?

A digital image provided by Trimble Image courtesy of Trimble

That’s certainly the view of Aviad Almagor, vice president of technology innovation at technology company Trimble, who predicted that so-called ‘agentic AI’ could start to play a major role in making construction workflows more efficient within the next year.

Almagor was speaking to Construction Briefing at the recent Trimble Dimensions event in Las Vegas, US, in November, an event that attracted over 7,000 people – a record number for the event. Many of those wanted to talk and hear about AI, with AI-related sessions oversubscribed and attendees often left to stand at the back of rooms as there were no free seats left in some of the AI sessions.

AI is nothing new for construction – or for Trimble – but over the last 18 months or so the whole world seems to have woken up to its vast potential. During a relaxed 30-minute conversation over a coffee, Almagor underscored how he expects the technology to redefine workflows, boost productivity, and enhance collaboration across the sector.

Agentic AI is a more advanced form of artificial intelligence designed to act with a degree of autonomy. Unlike traditional AI systems, which respond to direct user inputs, agentic AI can independently assess tasks, set priorities, and make decisions based on contextual understanding.

“Basically, it’s about humans providing an instruction – where we want to get to in this complex construction process – and letting the agent play with estimating, scheduling, material and so on to bring an optimised solution into our hands. That’s the future, which I think we’ll start to see at least portions of in the next 12 months or so,” Almagor said.

From predictive AI to agentic

AI is not new to construction and, as Almagor pointed out, Trimble has been integrating the technology in one form or another into its products for some time.

Predictive AI tools have already been applied to large masses of raw data (for example from a point cloud generated by a drone survey), which can be time-consuming to analyse, in order to turn it into useful information for decision makers.

But the industry is now evolving towards more advanced workflows enabled by generative and agentic AI.

“Generative AI brought a boost because, for the first time, users could interact with AI systems in a natural way,” Almagor explained.

Agentic AI, with its capacity for autonomy, promises another significant leap. By embedding artificial intelligence directly into workflows, it can take over some of the more mundane or time-consuming tasks, at the same time as making it an easier and more natural process for software users to learn their way around the system. In theory, it should mean less human intervention and a bump-up in productivity.

Almagor summed up three capabilities that Trimble will provide to agentic AI in order to make them autonomous and capture those productivity gains. They are:

  • The ability to reflect on what they are doing and iterate, based on knowledge that Trimble has loaded into the system.
  • The ability to plan ahead, defining priorities on a project and the tasks that need to follow.
  • Access to tools and real-time data from sensors or historical datasets that allows the AI agent to deal with more complex operations compared to simple or generative AI.

“Imagine we have several stakeholders in the industry: architect, engineer, general contractor, subcontractor, owner – and each one of them will be represented by an agent,” Almagor said.

“If you take architecture, for example, you might have sub-agents for the architect and one of them is an expert in installation processes, one in 3D modelling and scripting, one reviewing design processes, one looking at interoperability.

“Then let’s say you want to connect your tribe of agents with a third-party agent because you need to collaborate. All this will happen autonomously. The idea behind it is to provide autonomy to the system.”

Challenges and the Path Forward

Despite its promise, AI still faces challenges. First, there is the apprehension some feel towards using new technologies, “People are concerned about something they’re not familiar with,” Almagor said, adding that building confidence in AI systems is crucial.

Portrait image of Aviad Almagor Aviad Almagor (Image courtesy of Trimble)

Second is ensuring that governance is robust enough that users are confident to share potentially sensitive data.

For AI to deliver on its potential, it requires vast amounts of high-quality data, which often resides with customers.

“Some customers are concerned about what happens to this data and this is something we need to solve with them, to build trust and make sure they understand that as long as they are using their siloed datasets, they are very limited in what they can do with AI,” Almagor said.

“We collect data from thousands of customers and anonymise it and it trains our models. Critical to our ability and the industry’s ability to moving forward is getting the data right. It’s certainly an interesting challenge.”

To protect data, Trimble has two connected streams when it comes to developing AI, Almagor explained. On the one hand, there is an engineering stream where engineers explore and identify opportunities to develop new AI capabilities.

On the other is a governance board for AI, which sets internal and customer-facing policies that explain what Trimble is doing with the data and what it is responsible for. “Many of our AI-driven tools are on the Microsoft Azure AI platform, which provides the umbrella of a secure environment where we know the data we share is safe and secure,” he added.

A new era for construction?

Nonetheless, there is no one type of technology that excites Almagor more than others when it comes to their potential to bring change to the construction industry.

Instead, his enthusiasm stems from the convergence of multiple different technologies.

“If you bring together mixed reality and AI, for example, it’s not just about visualising digital content in the context of the physical environment. With its ability to interact between the two, AI can help you understand what you see and understand the relationship between physical and the digital,” Almagor said.

He added, “I’m pretty confident we’ll see AI in almost every Trimble product at different levels, from basic interaction to more complex agent workflows and multi-agent environments.”

Form placeholder

STAY CONNECTED

Receive the information you need when you need it through our world-leading magazines, newsletters and daily briefings.

Sign up

CONNECT WITH THE TEAM
Andy Brown Editor, Editorial, UK - Wadhurst Tel: +44 (0) 1892 786224 E-mail: [email protected]
Neil Gerrard Senior Editor, Editorial, UK - Wadhurst Tel: +44 (0) 7355 092 771 E-mail: [email protected]
Catrin Jones Editor, Editorial, UK – Wadhurst Tel: +44 (0) 791 2298 133 E-mail: [email protected]
Eleanor Shefford Brand Manager Tel: +44 (0) 1892 786 236 E-mail: [email protected]