Gord Rawlins, President & Chief Executive Officer — CMiC, recently shared his vision for how AI is revolutionizing the construction industry. Through strategic implementation of natural language processing, advanced analytics, generative AI, and process automation, CMiC's construction ERP platform is setting new innovation standards in project management.
Key features like "AI" and Project Pulse deliver real-time insights and automation capabilities that are transforming project management through advanced construction AI. This comprehensive Q&A explores how CMiC's platform supports digital transformation through project automation and efficient resource planning, enabling firms to drive greater efficiency and reshape the industry.
Rawlins covers essential topics including AI's role in construction, natural language processing applications, advanced analytics capabilities, and the transformative power of generative AI.
Q: What are your thoughts on AI impacting the construction industry?
A: AI is having a major impact across all industries, but I see unique and interesting applications in construction. Consider its use in design, scheduling, and robotics. I've seen work crews—drywall crews and bricklaying crews—all using robotics. There's autonomous equipment; I've seen an excavator dig a trench at exactly the right distance and depth with nobody operating it.
You have drones monitoring construction sites, capturing images and videos to track progress and identify safety alerts through visual analysis. There's wearable technology with visors that overlay augmented reality onto actual reality.
But honestly, none of that is what we're involved with. We do construction ERP software, so our AI focus centers on a completely different set of priorities. We love seeing those field developments and we interface with those products, but our product plan and roadmap focuses on four main AI categories: basic natural language processing, advanced analytics functionality throughout our system, generative AI for system-generated content, and process automation for standard workflows.
Q: In your perspective, what is the significance of natural language processing to the construction industry?
A: Natural language processing has led this wave of AI adoption over the past couple of years. AI has existed for decades, but it was in the realm of data scientists. With natural language processing, anyone can now utilize this technology.
I experienced this firsthand at a family dinner in late 2022, right after ChatGPT launched. My 16-year-old niece mentioned she had to work on a school essay. I asked about the topic and showed her ChatGPT—she'd never seen it before. It completely wrote her essay in literally seconds.
Her reaction made me realize the impact this would have on everything because you can just ask for something and receive it without knowing the correct syntax or how to frame a question. It's like speaking to a person. This natural language processing ability is the foundation of everything happening with AI today.
We started our initiative right after that experience. At the end of 2022, we assembled our team to develop our product called "Al." This enables users to query our ERP data with any question in natural language.
We're releasing this in final mode with our next version. People will be able to ask their system: "Give me a list of all jobs that didn't submit daily journals last night," or "Show me which jobs had low productivity levels last week," or "Based on current trends, what are my cost projections and which top 10 jobs have the highest chance of cost overruns?"
You can even do document analysis with “Al.” You can ask it to review posted change orders or any object in the system. Questions like "Did I get paid yet for that job?" or "Has that requested change order been sent?" Just ask in English, and the system provides the answer.
This ability completely revolutionizes how people communicate with their systems. They no longer need to learn how to ask questions or where to find answers—they just pose the question. This natural language query capability was our first real AI initiative, and everyone who's tested the beta and prototypes has been blown away by its impact.
Q: Can you elaborate more on "Advanced Analytics"?
A: Advanced analytics uses natural language but is interspersed throughout the entire system, performing different types of analysis. Our first implementation was "Project Pulse," which does sentiment analysis.
Sentiment analysis has AI review content written in daily journals, RFIs, meeting minutes, or issues, determining whether the sentiment is positive or negative based on word usage. If a job's overriding sentiment is negative, it might need immediate attention. Positive sentiment is obviously a good sign.
We have this feature in our analytics tool, already released in beta. Project Pulse displays a map showing job locations and their sentiment analysis rankings. It's been really interesting and moves quickly to the next level.
While sentiment analysis only reviews text-based entries, the next level is document review, examining anything throughout the system—whether in our enterprise content management system, imaging system with scanned documents, or any entered objects like subcontracts or change orders. It performs similar sentiment analysis but looks for risk items or anything requiring attention.
People don't want to spend time reading long text anymore. They want AI to review documents, summarize them, highlight key points, and indicate necessary actions. For compliance notifications when entering new countries, understanding regulations, or reviewing subcontract terms and conditions from new partners, AI can handle these reviews efficiently.
Anomaly detection is another advanced analytics function, looking for irregularities based on historical data. An AI engine might examine a job and say, "This doesn't look right—you haven't completed enough work based on hours spent," or "It looks like something was missed in this situation where you'd typically have something identified."
With anomaly detection, sometimes you don't know what question to ask. Unlike “Al” where users drive information requests, anomaly detection has the system examining everything and saying, "You might want to look at this—it doesn't look right."
Then there's predictive analytics or predictive modeling, where the system analyzes past patterns to predict future outcomes. We use this in budgeting, where we examine total job budgets for cost codes but can spread costs across periods based on similar historical jobs. It's no longer a manual task of schedule-based allocation—AI handles the first pass and spreads costs automatically, doing the same for projections.
Other advanced analytics include GPS monitoring for location tracking, image analysis from drones, and IoT monitoring on equipment for efficiency assessment and maintenance scheduling. These are all analytics that systems will perform automatically going forward.
Q: What will be the importance of Generative AI to the construction industry in the future?
A: Generative AI is the fourth major category impacting construction ERP. Systems have always generated things—reports, forms, documents, subcontracts, financial statements. But with generative AI, it's no longer up to users to decide how to build these things. Systems will create them automatically.
The system will understand what a form looks like and produce it. You'll be able to say, "Generate a balance sheet for this company as of this date," and it will do it because it knows the format and your data. We'll ensure it's trained on your data so it can produce outputs without requiring formatting.
This extends to reports. Currently, you have a menu of reports, and modifying them requires IT or software vendor involvement. With generative AI, users will be able to say, "Please print that same report but add the column for uncommitted costs," or "Please give me a subtotal every time the location changes," and it will regenerate the report automatically.
This is somewhat more futuristic than other items, but generative AI will have computers generate things that people currently create. In an ERP managing so many elements, the possibilities are endless.
Consider imaging with document types containing specific fields. When scanning images, you want to extract data into the database—that's OCR technology. With generative AI, you'll be able to say, "I want to take that data from that form and create a fly sheet"—a data entry screen mimicking all form fields. This fly sheet becomes an entry point where users can define appearance, add fields, set valid values, and publish for all users.
Previously, only developers could do this. In the future, anyone will have this capability, completely revolutionizing system interfaces.
Before leaving generative AI, I want to discuss process automation—our fourth category. Process automation and generative AI will generate workflows in our system. While systems have always supported comprehensive workflow processes, users will now have that control.
We're developing an AI-enabled Workflow Builder that allows natural language workflow creation. For example, you could type: "Let me know every time a cost is posted that causes the budget to be exceeded. Send me an alert every time that happens," or "Send me an alert every time a subcontractor indicates a deficiency or submits an RFI with cost impact. I want to see that immediately in my inbox."
These workflows were previously possible but required programming. Now you'll generate workflows just by speaking them.
Consider finance department processes like month-end procedures. You can tell the computer to close the month, set the current period to the next period, and execute. While there might be reversing entries requiring specific adjustments, eventually the system will handle these too.
Cost reallocation—distributing overhead costs to jobs based on hours or dollars spent—can be automated. You tell the system what you want; it executes without requiring setup knowledge. Eventually, you chain these together: post entries, close the month, set the next period, and produce all monthly financial statements. The system will handle everything.
Q: How will AI impact individuals performing tasks in the future?
A: In the future, more mundane tasks will be done by computers and fewer by people. This raises the question: "What happens to people if computers do everything?" But if you think about it, it means people working these jobs become managers, managing autonomous systems—essentially managing robots and telling computers what to do.
Systems don't know what to do unless directed. I use the analogy of a self-driving car: people still decide where to go—they just don't have to drive. The car drives, but you're still needed to decide the destination. Without human control, it's pointless.
Earlier I mentioned autonomous equipment—an excavator that dug a perfectly straight trench at exact depth. Previously, a person would have done this, but probably not as well, taking longer and not working as many hours. However, there was a person who created that video—a robotic process engineer controlling the equipment, determining where to dig, analyzing, and setting up safety barriers. The person manages the equipment, changing how they think about their job. They're no longer doing things systems can do—they're doing things requiring more brain power.
I'll share another example. In the early '90s, I went to Slovakia to implement a system for a telecommunications company. I was showing them automated workflows when someone asked through a translator, "Where do we enter the call records?"
This was a country just emerging from Communism. Many people entered data, including call records for call start and end times. When asked about the data entry screen, I explained they wouldn't need to do that anymore—the digital switch would provide all call records automatically.
There were about 50 people in the room, and the translator said, "I don't think you want me to tell them that. That's all these people do." I responded, "What are you going to do then? The switch does it. I don't think these people want completely unnecessary jobs." They said, "We're not prepared for what we'll tell them they'll be doing."
I said, "That's not really a problem. It's about eliminating mundane, unnecessary work. There must be something more valuable they could do instead."
Today, many years later, those people no longer do that work. Everything is digital and electronic. Data transfers electronically all the time. That entire job description from 30 years ago is gone.
This is what will happen with AI. AI will take jobs that are currently necessary because there's no other way to accomplish these tasks. But systems will handle all of that.
Jobs will change and become better, more fulfilling, and more productive because they'll be necessary rather than just creating work for people. AI will change the workforce and many other things, having people work in completely different ways, but it will be much better long-term because mundane tasks that can be done by machines will be done by machines.
The machine will never be able to decide what to do. People will always decide what to do; the machine will just execute it.
To learn more about CMiC’s AI-powered capabilities, please click here.