Turning years of data into smarter decisions with Agentic AI

The Challenge

Even with a centralized platform in place, finding the right information at the right time remained one of the biggest challenges on Haskell’s job sites. Teams were still spending valuable time digging through drawings, RFIs and submittals, slowing down decisions and adding friction to already complex workflows.

The Solution

By adopting Datagrid, Haskell layered AI-powered agents on top of its Procore data, allowing teams to search, analyze and act on project information instantly. Instead of navigating documents manually, users can ask questions in plain language and receive targeted, verifiable answers—unlocking the full value of the structured data already captured in Procore.

The Results

• Submittal reviews reduced from 1 week to 10 minutes
• Faster access to critical project information in the field
• Reduced administrative burden across project teams
• Improved decision speed and confidence
• Early adoption across multiple projects, including their HQ, with rapid buy-in

A simple but time-consuming problem: finding the right information

Haskell, a global design-build firm known for delivering complex industrial and infrastructure projects, has spent years standardizing how its teams capture and manage project data.

On paper, that should make finding information easy. In practice, it’s often anything but.

Answers are rarely far away on a construction site—but they’re often hard to find. For Haskell’s teams, that might mean walking from the field back to the trailer, opening a set of drawings, flipping through pages and still not finding what they need. Or pulling up documents on a tablet and searching through layers of files for a single detail.

“It’s not that the information isn’t there,” said Hamzah Shanbari, Director of Innovation at Haskell. “It’s that you’re looking for two words on one page in a massive set of documents. That’s the bottleneck. There’s a lot of wasted time and energy just trying to find the right information so you can make a decision.”

Even with Procore in place as a centralized system, that last step—finding the exact piece of information needed to move forward—remained frustratingly time-consuming.

A foundation built on structured data

Luckily, Haskell had already done something many companies struggle to achieve: it standardized how project data is captured across its organization. RFIs, submittals, drawings and daily logs are all captured in Procore in a structured, repeatable way, creating a reliable system of record across every project.

“If something doesn’t integrate with Procore, we’re not talking to them,” Shanbari said. “You have to have that unified, central platform. Otherwise everything becomes its own silo, and it’s a whole lot harder to get a read across the company and each project.”

That discipline created a powerful foundation, especially for a firm trying to operate consistently across a vast global footprint. But accessing that data still required manual effort.

A new layer: Datagrid

Datagrid changes that equation. Layered directly on top of Procore, Datagrid is an agentic AI platform purpose-built for construction that provides unified intelligence, making it easy for users to get verifiable, source-backed answers. Unlike generic or black box AI, its specialized agents go beyond simple search to both recommend and execute on work, transforming complex project documentation into immediate, high-confidence action

“Datagrid allows us to stop digging through documents,” said Ashley Sullivan, Innovation Strategist at Haskell. “Now we can ask a question and get the answer back in context. It’s not taking the human out of the equation—it’s helping them think through what they’re seeing and what they should do next.”

That shift is subtle, but significant. It doesn’t just make workflows faster—it changes how people approach their work. “It’s a completely different mindset,” Shanbari said. “You’re not searching anymore. You’re asking for what you need.”

Turning weeks into minutes

One of the clearest examples is submittal review. On a typical project, teams might spend days—or even a full week—reviewing submittals against specifications, identifying gaps and sending them back for revision.

With Datagrid, that process can happen in minutes.

“We took something that would normally take a week and did it in 10 minutes,” Shanbari said. By uploading a revise and resubmit submittal to Datagrid, Shanbari’s team can instantly receive a clear table detailing compliance issues. “It immediately tells you what doesn’t comply, what’s missing and what needs to be addressed. If you multiply that across hundreds of submittals on multiple projects, the impact is huge.”

By catching issues earlier, teams can avoid unnecessary delays and reduce the number of revision cycles. But turning those gains into day-to-day reality required teams to adopt a new way of working.

Construction teams aren’t always eager to adopt new technology—especially something as unfamiliar as AI. Haskell approached rollout with that reality in mind, starting with simple use cases and allowing teams to build confidence before expanding into more advanced applications.

“AI is very daunting,” Sullivan said. “You’ve got people who don’t know the first thing about it, and others who use it every day. Datagrid meets them where they are. It gives them something simple to start with, and then they can grow from there.”

The team began with basic search functionality, then gradually introduced more specialized agents as users became comfortable. “It only took a couple of weeks,” she said. “Once they trust the answers and can see where the information is coming from, they start asking what else it can do.”

Faster answers, better decisions

As adoption has grown, the biggest impact has been on decision-making. With faster access to relevant, up-to-date information teams can move forward with greater confidence–and less back-and-forth. For Haskell’s younger teammates in particular, Datagrid’s recommendations for next steps and followup actions removes the guesswork and allows them to think through problems with deeper understanding. 

“It’s about getting to decisions faster with high confidence,” Shanbari said. “The more we can reduce that cognitive load of trying to find the right information, the more efficient our teams become.” 

That applies across roles. Field teams use it to locate details quickly, while leadership relies on it for faster, higher-level insights. “Different people ask different kinds of questions,” he said. “But it’s all the same data. You just need a better way to get to it.”

While Haskell relies on many of Datagrid’s out-of-the-box agents—Deep Search, Submittal Reviewer and RFI Checker to name a few—it’s the ability to create custom agents that makes the platform such a difference-maker for the team. “Most AI tools are purpose-built to do one thing, so they're not giving the client the ability to tweak it or create their own methodologies and ways of getting to the information,” Shanbari said. “With Datagrid, the sky's the limit.”

And because Datagrid sits directly on top of Procore, it enhances existing workflows instead of replacing them. “You’re not pulling data into another silo,” Sullivan said. “You’re working directly with what’s already in Procore. That’s what makes it powerful.”

That integration also opens the door to broader use cases, from connecting additional data sources to automating repetitive tasks. “There are so many small things that add up to hours and hours,” she said. “If we can take some of that off people’s plates, they can focus on the work that really matters.”

A future of zero RFIs

For Haskell, the long-term vision is ambitious. “Zero RFIs—that’s what we’re shooting for,” Shanbari said. “If design is complete and accurate, and you can get the right information instantly, construction becomes so much smoother.”

It’s a future where fewer decisions are delayed, fewer questions go unanswered and teams spend less time searching and more time building.

“This is the future,” he said. “If you’re not going in that direction, you’re going to become irrelevant in five years or less.”