Fixing the inputs so AI can deliver the output

The Challenge

As Bernards managed dozens of complex projects simultaneously, project teams were burdened by a massive volume of unsorted data. Engineers were wasting time with tasks like scrubbing submittal registers or searching for contract parameters. The situation made it difficult to reconcile a team’s perception with the reality of project health. Without a way to automate these tasks, identifying risks often relied on gut feel or luck.

The Solution

Bernards standardised on Procore as a single source of truth for project data, replacing fragmented spreadsheets and manual processes with structured, consistent information across every job site. With that foundation in place, the company is deploying Procore AI to automate tasks like RFI and submittal review, helping teams identify risks and discrepancies while shifting their focus to high-value decision making.

The Results

• Established a single point of entry for project data, rendering it consistent and AI-ready
• Enabled real-time tracking of leading indicators like RFIs and buyout progress
• Identified risks using structured data and project controls
• Established automated QA/QC to catch data outliers and used AI agents for predictive risk mitigation
• Shifted to a coaching culture, using AI to bridge perception and reality while empowering teams to ask questions and learn without hesitation

Garbage in, garbage out: building an AI-ready foundation 

For Bernards, generating data from RFIs, schedules and cost updates was never the primary hurdle. The real challenge lay in ensuring that information was consistent enough to be actionable. Without standardised discipline, teams often relied on gut feel, causing critical issues to surface far later than they should. 

As a builder and civil engineer, Mandar Joshi, the Director of Project & Process Controls for Bernards,  recognised that moving the company forward required a foundation of data integrity. By standardising on Procore, Bernards established a single point of entry for every tool, allowing project teams to follow a consistent workflow that rendered data AI-ready from the start. 

“What Procore has really helped us do is establish a certain set of procedures and processes,” Joshi says. “There’s one point of entry to each tool, which helps ensure that everyone is following the same steps — because without those, you won’t get the result you’re looking for.” For Joshi, the goal is simple: “When you have good input, you get good output. You have garbage input, you get garbage output.” 

Bridging the gap between perception and reality

With structured data in place, Bernards has transitioned from reactive project management to predictive oversight. By tracking leading indicators — such as average RFI handling times — the company can identify "slippery slopes" before they impact the master schedule.  

This data-driven approach allows leadership to verify the qualitative stories coming from the field against quantitative reality. “The team’s perception is the qualitative side,” Joshi says. “What the data’s telling us is the quantitative side, and that’s reality. How do we bridge that gap?” By identifying risks "hundreds of miles away,” Bernards can allocate resources to address problems before they become crises. 

From search to strategic oversight

The deployment of Procore AI represents the next evolution in this strategy, shifting the team’s focus from document research to high-value decision-making. While AI simplifies search — allowing teams to find contract parameters or soil export details — Joshi’s vision centres on specialised agents that "swarm" complex project issues.  

“You have an RFI agent, a submittal agent, a schedule agent, and they talk to each other,” Joshi says. “They present a synopsis, and then you take that to a supervisor. That’s where the value is.” This automated ecosystem is designed to perform a three-way comparison between drawings, specifications and submittals, identifying discrepancies instantly that a human eye might miss.

Ending the “RTC” era: mentorship over administration

For Joshi, the most significant impact of layering AI onto structured data is the cultural sea of change it facilitates. When he started in the industry, asking a basic question often led to a dismissive response. “I remember asking simple questions and being told, ‘RTC’ — read the contract,” he says. “You kind of feel dumb, like you should have read it, but sometimes you just need a quick answer.”

With AI handling the tedious research, that dynamic is shifting. The ability to ask a question without hesitation has removed the harshness of the old culture, making it easier for teams to find guidelines without burdening superiors. However, Joshi is clear that the technology is a partner, not a replacement. “AI is something you should learn from, not completely depend on to do your tasks,” he says. “It provides the synopsis, but you take that to your human supervisor to make the final decision”.  

By automating the "busy work," Procore AI reclaims time for high-level coaching, letting the project leader act as a strategic mentor, not a manual data researcher. “When AI flags a risk, it facilitates a better conversation between a team member and their supervisor,” Joshi says. “It creates a moment for coaching and allows people to ask the right questions to make sense of the data.”