AI Wins This Week: From Call Prep to Business Intelligence in One Monday Afternoon
How I spent Monday afternoon "playing" with AI and accidentally built our most strategic business tool
I started Monday afternoon trying to save Kevin 30 minutes of discovery call prep. By 6pm, I had accidentally built a business development intelligence system that gives us strategic insights before the first conversation. Here's how a simple experiment evolved into a tool that may change how you think about what AI can do for your business.
The Spark: A Zapier Newsletter and Post-Lunch Playtime
After a morning filled with the usual financial work (ah, the glamour), I rewarded myself with some AI playtime. I'd recently read a Zapier newsletter about using Agents—a tool I was eager to master. While still learning, it got me thinking: Kevin spends a lot of time before each discovery call doing basic research, like checking property details and neighborhood context. What if we could automate this to save him time and increase accuracy?
Additionally, our LSU student intern would also spend another 15-30 minutes locating the project address on Google maps, Google earth and checking out real estate sites for existing images of the property.
My Monday afternoon goal: See how much of that I could automate.
What started as "let me play with this for an hour" turned into something that fundamentally changed how we evaluate projects and a day later was placed in office workflow as a "thing" we do.
The Technical Journey (Or: How I Built Around a Roadblock)
I dove into Zapier's Agent builder for the first time, dictating my workflow ideas as I went. Think of an "agent" as a more sophisticated automation—instead of just "when this happens, do that," agents can make decisions and handle complex, multi-step processes. Things were moving along really well until I needed the system to connect with Claude's API.
API what? It's essentially how different software programs talk to each other—like having a translator that lets Zapier communicate with Claude and pull research automatically. But Zapier's agent couldn't make that conversation happen.
Not one to take an AI no for an answer, I asked Claude for a workaround. His suggestion was refreshingly simple: since this process was linear, just build it as a regular Zap instead. Regular Zaps are Zapier's basic automation—simpler than agents, but perfectly capable for this job. Plus, he assured me that agents could still be in my future once I get more comfortable with the basics.
Here's what I built:
Trigger: New application comes into Dubsado (our CRM)
Research Phase: Zapier sends a detailed research request to ChatGPT with web browsing
Intelligence Phase: ChatGPT delivers comprehensive findings to Claude
Analysis Phase: Claude synthesizes everything into a professional brief
Delivery: Kevin gets a formatted email with complete prospect intelligence
The whole system was working within a couple of hours. But I couldn't stop there. It was clear it could do more - better.
The Iteration Reality
Here's what I learned that every AI builder needs to know: You're going to iterate and iterate and iterate. One amazing breakthrough begets another breakthrough. It's addicting.
My first version was perfectly acceptable—better than what we typically provide Kevin ahead of discovery calls. "Acceptable" became "what if we added this?" which became "holy cow, look what this can do."
I also discovered something I "knew" but didn't experience. Different ChatGPT models format responses completely differently - even with identical instructions. Same research request, totally different output quality depending on the selected model. This was my first time experimenting with the dropdown model options, and it was eye-opening for future automation and a game-changer for getting clean, usable results.
The Breakthrough Moments
Moment 1: The Google Earth Discovery
While demonstrating the system to our LSU student, I had an aha moment. We always ask her to pull Google Earth images for site visits. What if AI could generate the property context automatically? After a few tweaks, it started delivering Google Maps and Earth links, saving us time and ensuring Kevin had the right visuals for every consultation. Beyond just "seeing" the property, the Google Earth images often show rough outlines of property boundaries—quick information about whether there's room for an addition or if a variance might be needed.
Moment 2: The Strategic Shift
Once we saw how comprehensive this intelligence was becoming, I asked Kevin a critical question: "When in the process do you want this kind of research?"
His answer changed everything: "I want it as soon as possible."
That's when I moved the trigger from "discovery call scheduled" to "application received." We weren't just preparing for calls anymore—we were doing business development intelligence that helps Kevin decide whether prospects are worth pursuing at all.
What We Actually Built: Business Intelligence, Not Just Call Prep
The final system delivers:
Property Intelligence: Market value, price per square foot, neighborhood comps, zoning context
Project Assessment: Budget-to-scope considerations and potential opportunities
Visual Intelligence: Google Earth and Maps links for immediate property context
The result: What used to take an hour of manual research is now done in 5 minutes. Not only that, Kevin now has strategic insights he didn't even know he needed before talking to a prospect. This is how AI isn't just saving time—it's helping us make better, more informed decisions from the get-go.
The strategic advantage: While most architects are winging it, Kevin has comprehensive intelligence on every prospect before they even talk.
The "Of Course" Moment
By 6pm, when we tested the full system, Kevin's reaction was perfect: "Of course. Of course this is what we needed." He was blown away—he had everything he could want right there.
That's the beauty of great automation—once you've experienced it, it's hard to imagine going back to the old way. But more than that, it often reveals possibilities you couldn't see before.
We'd accidentally built something that transforms how we evaluate opportunities, position our services, and engage with prospects. What started as "let's save 30 minutes" became "let's make strategic decisions based on real intelligence."
What This Means for Your Business
The lesson isn't about the specific tools (though Dubsado + Zapier + ChatGPT + Claude + email is a powerful combination). It's about what becomes possible when you start small and stay curious.
Three things to try:
Pick one manual task that takes 30-60 minutes of your time regularly
Start experimenting without knowing exactly where it'll lead
Ask "what else could this do?" once you get something working
The compound effect: Small automations become strategic advantages faster than you think. What feels like "just playing around" can become the thing that changes how your business operates. This isn't just about efficiency—it's about freeing up Kevin's time for things that only Kevin can do, reducing decision fatigue, and enabling faster, better-informed decisions without delays waiting for him to have time to focus on each inquiry.
Combined, our business development and pre-call prep that used to be a one-hour task is now 5 minutes, completely automated, and delivered timely to Kevin's inbox with the right label.
What task in your business feels like a time drain? Have you tried automating it? Sometimes, small experiments lead to big changes—give it a try and see what you can build with just a little AI tinkering.
Elizabeth Harris handles AI integration, business operations, and occasional late-night automation experiments for Kevin Harris Architect. When she's not accidentally building strategic business tools, she's asking AI which of her eyeglasses work best with her outfits and which two pair to pack. She experiments with every AI tool she can get her hands on—including using them to enhance her writing process—but believes no tool can replace authentic personal experience and genuine curiosity. Subscribe for more insights on using AI to work smarter, not harder—no computer science degree required.