The first AI Industry Night had no slide deck, which was the point.

A room full of people showed up at Codefi's Cape Hub on Wednesday, September 2, and the hour ran entirely on their questions. Cara Dirnberger, Codefi's Director of AI Programs and Workforce Innovation, opened by naming the reason the series exists: to get people from across southeast Missouri into one room and work out how AI actually fits into their day-to-day jobs. Then she handed the floor to Brian Holdman and Garrett Duncan and let the room drive.

Brian is Codefi's CTO and has been writing software for more than 30 years. Garrett is a software developer who, by his own description, lives in AI around the clock. Between them they fielded questions from a home renovation contractor, an insurance professional who also manages a farm, a founder building a routing app for touring musicians, and several people who had opened ChatGPT once and weren't sure what to do next.

Start with the part of your job you don't like

Most people come into a room like this wanting to know what AI can do that's impressive. The answer that got the most agreement all night pointed the other direction. Someone asked whether there were business tasks that were just no-brainers to hand off — and both Brian and Garrett landed in the same place. Find the repetitive one.

"Look at AI to take those mundane recurring items and tasks that you just don't really enjoy," Brian said, "so you can focus on the parts of your job that you really like."

He had an example from that same afternoon. His team needed to dig through a large batch of emails, pull out specific ones, and compile them into a spreadsheet. He handed it to AI and had it back in five minutes. Done by hand, it was an hour of work.

That framing — start with the tedious thing, not the impressive thing — came up repeatedly. Garrett's addition was that once you've cleared the repetitive work, the next step isn't a bigger AI; it's a small tool built for exactly one person's workflow. "You can go build this tool that fixes that repetitive task," he said, describing what he now spends much of his time doing for local businesses. Not a product for the market. A thing that solves your problem, on your setup.

Treat it like an intern

The most useful mental model of the night came when someone asked how to stop getting mediocre results.

Brian's answer: think of AI as an intern. Ask an intern to go do something with no instructions and you'll get something back, but it probably won't be good. Give that same intern a standard operating procedure — exactly how you want this done, every time — and the work comes back solid.

That, he explained, is essentially what a "skill" is: a written procedure you hand the AI so it performs the task your way rather than guessing. Garrett described how to build one without any technical background. Keep working with the AI until you get the result you actually want, then tell it to save that process as a reusable skill. Next time, you call the skill and drop in your new information.

If it misses, you don't start over. You tell it what went wrong and update the skill. "You'll just keep reusing it at that point," Garrett said.

The example that landed hardest

A home renovation contractor asked about the time he loses to scheduling labor, estimating projects, and tracking materials.

Brian recognized the problem from his own consulting work — hours of client meetings, follow-up calls to clarify what got missed, then days spent writing proposals and quotes. His fix was an AI notetaker running in the meeting, feeding the transcript into a skill loaded with his own rates and past quotes.

The result: a proposal better than what he used to write himself, in five to ten minutes, roughly 95% finished.

For an in-person walkthrough of someone's kitchen, he suggested a phone recording app captures the same thing. Feed it in, add your pricing, and let it draft. It won't be perfect. It will save hours.

Brian also floated an idea nobody in the room had tried: photograph a client's kitchen before the job, tell AI to render it with the new countertops and cabinet color, and show the customer what they're buying in five minutes.

Which tool, and how to get better at it

An insurance professional asked whether to use ChatGPT or Gemini. Garrett's read of the current field: Gemini is the easiest to get to, especially on an Android phone, and the strongest for video. ChatGPT is the best fit for most business users and handles images well. Claude is strongest for data-heavy work, but costs more and hits usage limits faster. His practical advice was that ChatGPT and Claude are close enough that if you're comfortable in one, you're fine.

The bigger lever isn't the tool. It's context. One attendee noticed that AI seems to reward people who understand how the pieces of their own work fit together, and asked whether that was real. Garrett confirmed it: ask a generic question, get a generic answer, because the model is built to answer everyone in the room at once. Load in what you know about your own industry and the output climbs accordingly.

He added one warning worth repeating. AI tends to agree with you. Early on, that's the thing to watch for.

Vague prompts cause the other failure mode. "It tries to fill in the holes itself and makes a bunch of stuff up," Brian said, describing hallucination. Both recommended correcting early and often rather than letting a long answer drift, and telling the AI plainly not to go beyond what you asked for.

Next: real estate, October 7

The next AI Industry Night covers real estate on Tuesday, October 7. 

Cara made a point worth passing along: come even if you're not in real estate. Every session surfaces tools and methods that carry across industries, and the field moves fast enough that a month is a long time.

Also open now at the Cape Hub:

• Small Business Series — session three lands September 8, 12:00–1:00 PM, on marketing: social content and image generation. Garrett leads it.

• Founder Series — a new run of sessions for founders and aspiring founders, covering legal, accounting, and business planning.

• AI Builders Lab — office space at the Cape Hub for people actively building with AI, with mentorship included. Apply by emailing coworking@codefiworks.com

Learn fast. Build right. Gain traction.