Thiago da Costa of Procore: 5 Questions

In May, construction technology giant Procore announced it had introduced a new suite of agentic artificial intelligence to expand its Procore AI platform.

The added AI is powered by Datagrid Intelligence, a company Procore acquired in January, and which it has now integrated into its tech platform.

The expanded platform introduces AI capabilities for high-friction construction workflows, including submittals, requests for information, contract reviews, and searches across drawings and specifications. The added AI is designed to execute work directly inside Procore using new “actions” and “triggers” capabilities that can coordinate workflows and respond automatically to project events.

Thiago da Costa, senior vice president of AI and data at Procore, and former founder and CEO of Datagrid, spoke with Commercial Observer in May about Datagrid’s integration and what it means for Procore’s construction clients.

This interview has been edited for length and clarity.

Commercial Observer: Procore has added an agentic AI to its platform with the acquisition of Datagrid. What does that mean for clients, practically? 

Tiago da Costa: Taking a step back, Procore is the dominant record and system of action in construction. The majority of general contractors leverage Procore to build and to execute projects. What we are seeing with the advent of AI is how do we get rid of repetitive work people need to be doing that is also complex and requires them to think through that work. Engineers, project managers, all those folks are working inside Procore. They’re trying to cross-reference the documents and execute the project, making decisions where every decision has a huge impact on financial, labor, cost, schedule and so on. 

So, what we’re trying to do for our builders is to bring the best AI solution into that.

Since the acquisition, we worked in record time to bring state-of-the-art AI technology and workflows inside the Procore environment, because the majority of the builders are inside Procore. They make all their decisions there working every day inside that platform. An agentic AI does the work in the background for that user. So you can ask it to do something, come back three minutes later, and it has done the work for you. 

The goal is to remove repetitive manual work.

How is this different from what Procore had been providing? And how do you differentiate from your competitors?

Procore took an early start in using AI. They’ve been talking about AI and building AI capabilities since, I think, 2023, when ChatGPT and all these things came out. The company has been building on that, and last year they announced a bigger investment in AI at Groundbreak, their main user conference. 

With the acquisition of Datagrid, we merged all these technologies and capabilities together and converged into a singular experience. What they had initially was the ability to use AI to search and find documents. That was the key capability. When we built Datagrid, it was agentic. The difference is the AI takes over the work. If you’re doing research for Procore, the agentic AI does the work, takes the results of a search and keeps on working. 

So, if you say, “Hey, I need to review the submittal,” it’s a pretty complex process. It can take 10 hours for a project engineer to review 100 or 200 pages of a submittal against all the specs, which is another 3,000 pages. Imagine the AI taking over that task: doing the search, understanding those documents, and reasoning through the entire complexity of construction.  That’s a bit different than what competitors are doing, as most of them are working with text, because large language models and AIs are 100 percent product market fit with text generation utility. 

But construction is a visual industry. What we do is build models that are capable of dealing with visual problems — you have to understand points, the specs, a photograph, and you have to be able to combine all of that meaning, logic and decision-making that is coherent with how people work. 

Does Datagrid do this just for documents, or does it do it on site too, with cameras and other visualization?

Millions of photographs are integrated into Procore, and many of our customers take photos on their job sites and use our photos module. We can use the AI to actually know everything about the visuals on the job, photographs and even videos. It’s a critical capability of what we call multimodal intelligence. And we index all that data in a way that AI can actually use it.

We hear constantly that construction costs are going up and that technology is supposed to bring costs down, but there is a gap in the process — there isn’t enough skilled labor for actual building. How does this platform fit into the equation?

That’s a great question. There is a lack of skilled labor for work, and that is something AI won’t really do. We’re not going to get more plumbers, right? 

But what AI actually does is level up everyone faster. We bring together the best knowledge, index all of that data, and we bring that knowledge to the fingertips of anyone on the job that has the right permissions. And we bring that information to folks that are just getting started. It makes the person that just got started be an expert on something in a much shorter amount of time, because the AI can guide you.

Now that you’re integrated onto the Procore platform, what are your challenges as far as getting in front of clients and having them understand and accept what Datagrid is adding?

That’s an exciting question. A significant part of the Datagrid customer base was already on Procore. Our customers were raving about the acquisition because they’re very excited we’re going to work more closely together. We have a massive amount of demand for AI. We feel we have the right solution and that we’re a dominant player in the space. 

We believe that what really changed is we’ve simplified everything that had to do with permission, security, all the things that people really care about. So it really simplified how our customers think about adopting AI. They’re going just much faster now. The user is inside Procore, they want the technology to work for them, and they want the technology to be invisible.

Philip Russo can be reached at prusso@commercialobserver.com.