Your Business Already Has an AI Department. You Just Haven't Hired It Yet.
Right now, on your laptop, you can install a program that reads your files, writes emails, builds spreadsheets, runs reports, and executes tasks on a schedule while you sleep. No IT department. No enterprise contract. No six-figure software deal. An open-source tool called OpenClaw, a command-line agent called Claude Code, a handful of others. Free or close to it. Available today.
Most business owners have no idea.
I’ve spent the last year building AI automation systems for my own businesses. Not chatbots. Not “ask it a question and get an answer.” Actual systems that run daily operations. A morning briefing that pulls financial data, project status, and priorities into a summary before I open my laptop. A cost tracker that logs every dollar spent on AI and tells me where the ROI is. Custom skills that take an address, pull comps, run analysis, and produce an institutional-quality asset management report in two minutes. Things that used to take a junior analyst a full day.
I built all of this on my own machine. None of my data leaves my environment. There’s no monthly SaaS bill for the automation layer. I pay for the AI model’s compute the same way you pay for electricity. Use more, pay more. Use less, pay less.
The tools to do this are sitting in the open. So why isn’t everyone doing it?
The gap between available and usable is enormous
I watched my dad’s generation go through this with computers. He’s a homebuilder. Mid-90s, computers showed up in the office. Estimating software. Accounting packages. Email. The tools were available. Most builders kept running everything on paper and yellow legal pads for another decade.
It wasn’t stubbornness. They didn’t know what to automate. They didn’t know which software to pick. They didn’t trust that the machine would get it right. And nobody showed them how to integrate it into the way they already worked.
The ones who figured it out early had a real advantage. Faster bids, tighter margins, better records. The rest eventually caught up, usually by hiring someone to set it all up for them.
We’re in the same moment with AI agents. The tools are available. The adoption curve hasn’t started.
What an AI agent actually does for a business
Forget the sci-fi version. An AI agent is software that reads instructions you wrote, accesses the tools you give it, and does work. Repetitive work. The kind that eats your Tuesday morning every week.
Pulling data from three different sources and merging it into one report. Formatting invoices. Summarizing meeting notes and sending follow-ups. Monitoring a folder for new files and processing them automatically. Drafting responses to common client questions using your voice and your standards.
You write the instructions once. The agent runs them on a schedule or on demand. You review the output. Over time, you refine the instructions and the output gets better. That’s the whole model.
One agent handling one task saves you an hour a week. Five agents handling five tasks across your team starts replacing a part-time hire. An orchestrated system where multiple agents coordinate, one scoping the work, one executing, one reviewing the output against quality standards, that replaces a workflow that currently requires three people touching a spreadsheet.
My great-grandfather bought a tractor
Ty Ty, Georgia. Outside Tifton, where the roads are still dirt at the tri-county line. Early 1900s, my great-grandfather bought one of the first tractors in the state. My grandmother still has the purchase certificate.
His family thought he’d lost his mind. Cousins told him he was crazy. The horses work fine. They plow all day. They don’t break down. Why would you spend money on a machine that might sit in a field rusting?
He used those tractors to build one of the largest privately owned farms in the area at the time. The cousins kept their horses.
I’ve watched this cycle play out my entire career. The phone, the internet, email. My first job in institutional real estate, we printed paper packets for every meeting. Within months we banned printing entirely and everyone brought laptops. Then COVID forced us out of the office altogether. Management had been experimenting with hybrid schedules for employees with long commutes, but the real resistance was trust. If you weren’t visible at your desk, how did anyone know you were working?
Same fear every time. Same outcome every time. The tool wins. The people who adopted early had the advantage. The ones who held back spent the next cycle catching up.
AI agents are the next version of this. Your current workflows are the horses. They work. They plow all day. But the tractor is sitting in the lot with the keys in it, and the question isn’t whether your industry adopts it. The question is whether you’re early or late.
The fear is real and mostly wrong
I hear two versions of the same anxiety from business owners.
The first: “AI is going to replace my employees.” For most small and mid-size businesses, no. Your employees spend 30-40% of their time on work that’s repetitive, structured, and low-judgment. Data entry. Report formatting. Status updates. Follow-up emails. AI agents take that off their plate so they can spend time on the work that actually requires a human. Client relationships. Judgment calls. Creative problem-solving.
The second: “I don’t understand this and I don’t have time to learn.” Fair. You also didn’t have time to learn how to configure a CRM or set up QuickBooks from scratch. You hired someone, or your accountant set it up, or you bought the implementation package. Same model applies here.
What the DIY path actually looks like
You could do this yourself. The tools are open-source or cheap. OpenClaw installs with one command. Claude Code installs with one command. The documentation exists.
You’d need to learn how to write structured prompts that produce reliable output. You’d need to figure out which of your workflows are worth automating and which aren’t. You’d need to manage API keys, cost tracking, scheduling, and logging. You’d need to train your team on how to use it, modify it, and troubleshoot when something breaks.
For one person who’s technically comfortable, that’s a weekend project to get one agent running. Getting a full system deployed across a team with quality controls, role-based configurations, and cost visibility? That’s weeks of work and a lot of trial and error.
The alternative
Or you bring in someone who’s already built the system, tested it, and deployed it. Someone who shows up, maps your workflows, identifies the five things that will save you the most time, builds the automations, trains your team, and hands you a working system in weeks instead of months.
That’s what I do now. I take the infrastructure I built for my own businesses and deploy it for others. The framework is the same. The customization is specific to your operations, your tools, your team, your workflows.
Discovery takes a few hours. We map what you do, find the bottlenecks, and score each workflow on time cost and automation potential. Setup is a day. Build is a week or two. Training is hands-on, not a PDF you’ll never read. And if you want ongoing support, I’m there monthly to tune, expand, and troubleshoot.
My great-grandfather’s cousins eventually bought tractors too. Everyone does. The question is whether you’re the one who sees it early and builds while the field is open, or the one who catches up later after the advantage has passed. The tools are free. The expertise to deploy them is not. And the window is open right now.
Kris Van Meter is the founder of The Upland Group, an Atlanta-based firm focused on AI-powered business operations and single-family residential investment. He spent a decade managing institutional SFR portfolios at Sylvan Road Capital and Darwin Homes before building his own AI automation infrastructure from scratch.
New pieces land here first
One email when something publishes. No drip sequence, no reselling your address.