AI for small business: the unglamorous place to start

The best place to start with AI in a small business is not a strategy session — it’s the call that rang at 7 PM and went to voicemail, and the lead from last Tuesday that nobody followed up with. Fix those two things in your first 30 days. Everything else can wait.

Why unglamorous problems pay the fastest

Every founder I talk to wants to discuss AI agents, large language models, and workflow orchestration. That’s fine — eventually. But the businesses bleeding the most money are usually bleeding it in the most boring places: a phone that doesn’t get answered after hours, a CRM full of contacts that nobody has touched in weeks, a quote that went out and was never followed up on.

These aren’t technology problems. They’re process gaps that technology can now close cheaply and quickly. The AI part is almost incidental. What matters is that the gap closes.

Before you design any system, run this simple model on your own numbers:

Leak point	Your inputs	What to estimate
Unanswered calls	Calls/week × % going unanswered	Leads lost per week
Unsent follow-ups	Open quotes or leads × close rate	Revenue left on the table
Delayed response	Hours to first reply	Leads lost to faster competitors

Plug your own figures in. The output usually makes the priority obvious.

What does “fix the phone” actually mean in 30 days?

It means every inbound call gets answered — not by you, not by a part-time receptionist, but by a voice agent that can collect caller information, answer common questions, and book an appointment or route the call appropriately. This is not science fiction. It’s available today, and Business Runner is the platform I built specifically for service businesses that need exactly this.

The setup timeline is short. A basic voice agent can be live in days, not months. The configuration work is mostly scripting: what does the agent say, what questions does it ask, where does it send the information. If you want to go deeper on what these agents can and cannot realistically do, What AI voice agents can and cannot do (told straight) is the honest breakdown.

The metric you track in week one: how many calls came in, how many were answered by the agent, how many resulted in a booked appointment or a captured lead. Compare that to the week before. That’s your baseline.

How do you fix follow-up without hiring someone?

Automated follow-up is the second leak, and it’s often larger than the phone gap once you do the math. A lead comes in, gets a quote, and then silence. The owner meant to follow up. It didn’t happen. The customer hired someone else.

The fix is a simple sequence: lead enters your system → automated message goes out within minutes → if no reply, a second message goes out in a day or two → if still no reply, a third. The sequence stops when the lead replies or books. No human needs to manage this. Follow-up that never forgets: automating the CRM you already ignore walks through exactly how to set this up without rebuilding your entire stack.

The tools required: your existing CRM (or a simple one), a messaging platform (email, SMS, or both), and an automation layer to connect them. The sequence itself takes an afternoon to write. The discipline is in leaving it running instead of turning it off the first time it sends a message you would have worded differently.

In the businesses I run — including a real-estate brand operated end-to-end on automation — the two highest-return starting points, as of August 2026, have consistently been inbound call coverage and lead follow-up sequencing. Not because they are sophisticated, but because the gap between “zero automation” and “basic automation” in these two areas is where the largest volume of winnable revenue was sitting uncaptured. Both can be live within 30 days. Both produce feedback you can read in your own numbers within weeks. Every more complex layer of AI I’ve added came after these two were stable and measurable.

What does a realistic 30-day plan look like?

Here’s the sequence I’d run for a service business starting from scratch:

Week 1: Audit the leaks. Count unanswered calls. Count open leads with no follow-up. Run the revenue model on your own numbers. Pick the bigger leak first.

Week 2: Deploy the fix for leak one. If it’s the phone, get a voice agent live on your main line. If it’s follow-up, build the first three-step sequence in your CRM. Don’t perfect it — ship it.

Week 3: Measure. How many calls answered? How many follow-ups sent? How many replies or bookings generated? Adjust the script or the sequence based on what you see, not what you assumed.

Week 4: Close leak two. Now that one system is stable, add the second. By day 30, both are running. You have baseline data. You know what’s working.

This is not glamorous. It is also not complicated. The businesses that get stuck are the ones that skip this phase and jump straight to building an AI strategy. Which tasks should you automate first? A ruthless prioritization method gives you the framework for what comes after these first two.

When should you bring in outside help?

If you can configure a CRM and write a simple email sequence, you can do weeks one through four yourself. If you’re already stretched thin as an operator, or if the phone system integration is more complex than a simple swap, that’s when outside expertise pays for itself. A Fractional Chief Automation Officer can compress this 30-day plan significantly and make sure the systems you build in month one don’t create automation debt you’re cleaning up in month six.

Start small. Start where the money leaks. Build from evidence, not enthusiasm.

Want to talk through where your business leaks the most? The voice agent on this site is live — ask it anything.