Last month I had the opportunity to present "AI on the Jobsite" at ChippCon 2025 in Fargo, where industry professionals, consultants, investors, and AI enthusiasts gathered to explore practical applications for artificial intelligence. While initially I had some nerves about going to a tech conference as a carpenter, I soon felt at home among the community of ambitious builders from across the industry spectrum.

In the presentation, I shared how I'm using Chipp to help solve two of construction’s biggest problems - communication breakdown and inefficient execution. The tools I have built focus on issues that are well known in the industry. Problem #1: The top client complaint is that contractors are hard to reach, and the top complaint from other builders is that we aren’t bringing enough clients in the door. The presence of a customer facing website chat along with user data on the backend, can actually boost our ability to convene with clients that are qualified and give them top-tier human level support. Problem #2: Whether it’s double checking a contract, installing a new WRB, verifying a building code, or fixing a piece of equipment, accessing the information we need on-site is always a challenge. Increasingly I am moving away from scrolling through PDFs on my phone or watching YouTube videos, to simply having a conversation with the manual I need information from. Manuals you can chat with embedded as apps on your phone or with scannable NFC cards on each piece of equipment; these are just a couple of ways to access these resources easily.

One of the biggest takeaways from ChippCon was the importance of solving specific problems with AI, rather than just throwing it at everything all at once. While there are constant improvements in the tech that require us to adapt, the biggest hurdle right now is change management in an industry that can be slow to adopt.

Anthropic launched Claude's conversational voice mode in late May, and it's the first time I have actually felt this type of interaction to be useful. I know lots of contractors who use voice-to-text when they are onsite or driving around. With voice AI, instead of typing complex queries, you can simply speak naturally and get detailed responses. For folks experienced with the maddening lack of utility of Apple’s Siri, you may think this will just be more of the same. But honestly, it works so much better and it has only been available for about a month.

The recent addition of Claude Connectors is where the value really ramps up. Voice-mode aside, this integration is powerful however you decide to use it. Users can now connect Claude directly with their Google Workspace and Gmail. This means you can use text or voice to query project schedules from Google Calendar, dictate responses to subcontractor emails, and review specifications from Google Docs.

This could be incredibly powerful for contractors and design teams alike. With the current rate of change, I think it’s safe to assume that Claude will soon be able to speak numerous languages (only English right now) much like it can in text form. Despite being a natural skeptic, perhaps especially when it comes to construction tools, I am really excited to see where this technology takes us.

AI Agents have officially moved from concept to measurable performance in the construction sector. Turner Construction processes 30,000 contracts annually through AI agents, while Suffolk Construction's predictive safety system achieved a 28% reduction in recordable incidents. Procore's AI Agents reached general availability this year, managing RFIs, scheduling, and submittals autonomously. These examples go further than simply using AI to help write better emails, these are genuine problems being solved and mundane tasks being automated. This news is exciting for some, however…

The first thing I notice is these tools are still almost entirely focused on large enterprise construction firms. Secondly, implementation is going to be the biggest hurdle for most companies, regardless of size. The most effective AI projects embed it into existing workflows without requiring entirely new platforms. While large enterprises might be champing at the bit to embed AI everywhere, it typically takes longer to implement at these big firms. Whereas smaller contractors and design-build firms are able to move much more quickly, but only if they understand their options and see tools in the market they can actually use.

Three questions I think we should be asking when it comes to AI and the use of agents:

• What are my biggest time-wasters? - Maybe it’s educating prospects over the phone who don’t end up being a good client fit, maybe it’s following up with subs.

• What am I spending the most money on? - It might be marketing or manual data entry or staff meetings.

• What have we always wanted to do but haven’t because of either time or money? - Perhaps you have always wanted to offer more educational content to your customers or maybe you want to explore an entirely new market.

Start with these questions as a way of exploring where you might begin with AI in your business. While I am very interested to see where these tools go, my fear is that for big industry players, like Procore, they will be tempted to bog down their platforms with too many shiny new features - which often leads contractors to skip over them entirely.

Simplicity, simplicity, simplicity.

Despite these promising results, the Ramp AI Index reveals something not all that surprising: construction's dramatic AI lag. While 35.5% of U.S. businesses use AI as of Spring 2025, construction sits at just 4% adoption. Manufacturing achieved 12% adoption with documented productivity gains.

This gap represents opportunity. Since many of the current AI tools are being built for large-scale enterprises, those are the businesses being reported on. It can be difficult to understand where high-performance contractors or boutique design-build firms fit into this landscape. I try my best to infer from these broader trends and then seek out solutions that can be paired down and tailored to these more specialized industries.

Understanding both the potential and the implementation realities, I've been developing practical AI tools that address specific construction pain points:

HVAC Specification Assistant: This tool uses PHIUS standards and product catalogs from leadings brands like Zehnder as knowledge sources, providing instant answers about equipment selection, sizing calculations, and compliance requirements. Instead of hunting through multiple PDFs and spec sheets, design teams can ask natural language questions and get precise, cited responses.

Change Order Tracking via Text or Voice: This chat assistant can take disorganized text (and soon voice) information and turn it into change orders. The AI formats details into professional change orders and emails documentation automatically to builder and client. This could help eliminate the pain experienced when a client’s bill comes due and there is no paper trail for changes that were made to the original bid. Poor tracking of change orders is one of the most persistent yet solvable problems I see in our industry.

Weekly Site Log Automation: Project managers can type weekly progress updates conversationally, with AI automatically structuring them into professional reports for clients. The system enables all of your PMs to more easily keep their clients updated, while maintaining consistent formatting and professionalism. As a former project manager, I would typically leave this task until I was exhausted at home on a Friday evening, others might simply forget entirely. Creating less friction for manual processes like these can allow our teams to enjoy their role more while improving the way they serve clients.

These tools focus on eliminating administrative overhead rather than replacing expertise. AI is not a hammer, and not everything is a nail.

I’d love to get these tools in people’s hands and am currently looking for folks to test them. Shoot me a message if you’re interested in trying one out.