Wait, am I using AI? Should I? What AI can actually look like in your business.

There's a certain type of AI content that goes like this: "AI can help with marketing! And operations! And customer service! And fourteen other things!" And then it ends. No map. No next step. Just a vague sense that you should be doing more, somehow, with something.

Super helpful. Thanks, internet.

Most people start too big, or too vague, or on exactly the wrong task. Maybe they ask AI to do something that requires context it doesn't have, get a mediocre result, and use that to decide “Nope, not for me”. That's like deciding you hate cooking because your first attempt was a soufflé.

So before we get into what AI integration looks like across a whole business (and it can touch a lot more than most people expect) let's agree on one thing: the goal isn't to transform your operation overnight. The goal is to find one place where it actually helps first, feel that work, and build from there.

That's it. That's how to get started.

 The full picture: where AI can actually show up

Most small businesses and nonprofits have four or five categories of work running at any given time. AI doesn't land in just one of them …it has potential across all of them. Or maybe just some of them. Here's what that actually looks like.

Getting found and staying visible

This is marketing in the practical sense: keeping your business or organization in front of the people who need you. For most small operators, it's the first thing to slip when things get busy. There's always a more urgent fire.

AI doesn't run your marketing for you. But it's genuinely good at the parts that eat time without requiring your expertise: drafting social posts, writing Google Business updates, turning a completed job or program into a short story someone might actually read. Here’s an example: Mo runs an art studio in DC. Every piece he makes has a real backstory — the client, the materials, the idea that took three tries to land. He tried AI once for Instagram, got back something that read like a catalog, and gave up. Second attempt, he gave it the actual story. What came back was specific, a little funny, nothing corporate. He now batches posts on Friday afternoons with a refreshing drink and considers it the most enjoyable admin he does.

The same principle applies in nonprofits: program updates, donor newsletters, impact summaries, event announcements. The content is real…you just need help getting it out the door consistently.

Winning the work — or the funding

Estimates, proposals, grant narratives, quotes, scopes of work. The documents that sit between you and the revenue. These are almost universally good candidates for AI assistance, because they're high-stakes, time-consuming, and structurally identical every time you do them.

Walter runs a three-person HVAC company in Jackson. He estimates five to eight jobs a week. Each one takes forty-five minutes to write up, same structure every time, because apparently the blank document doesn't remember that he's done this before. He now gives AI the job details, his rate, and two things that make his company worth hiring. Draft back in under three minutes. He fixes the numbers, makes sure it sounds like him, sends it. Two hours a week returned to his calendar. He didn't change how he prices. Just how long it takes to write it down.

For nonprofits, this is the grant narrative. Michelle's organization has two talented grant writers who were spending the majority of their time reformatting similar reports for four different donors. It’s a similar story, similar impact measurements, and slightly different word limits and requirements. After the human thought goes into the report, the core activity is assembly. AI handles the first draft of the narrative sections now. Her grant writers spend their time on strategy, donor relationships, and the parts that actually require them. Turns out that's what they wanted to be doing anyway.

Staying in touch after the job is done

The follow-up gap is one of the most common and invisible revenue leaks in small business. The customer who loved the work and would have referred three neighbors — if anyone had asked. The donor who gave last year and never heard from the organization again until the next ask arrived. The consulting client whose project wrapped six months ago and who has probably forgotten your name.

Nobody means to let these slip. There's just always another job, another proposal, another urgent thing. AI doesn't close this gap by caring more than you do. It closes it by making the actual sending faster and easier. A follow-up text after a job. A check-in email thirty days after a project closes. A donor acknowledgment that sounds like a person wrote it. These take two minutes to draft with AI. They take twenty without it, which is why they don't happen. 

The back office: the stuff nobody sees but everyone feels

Scheduling, onboarding, intake forms, staff communications, meeting notes, internal SOPs that live only in someone's head. This is where small businesses lose enormous amounts of time— not on any single task, but on the accumulated friction of processes that were never properly built…and they’re usually used to doing business that way!

Priya is a solo accountant in Colorado. She onboards about 15 new clients a year and every year she has the same intake conversation, takes notes in three different places, and sends a welcome email she rewrites by looking in her emails to copy and adjust the last one. She got started using AI by spending two hours with AI building an intake prompt and a welcome email template. That was four months ago. Every new client since has gotten the same organized, warm first impression, whether she’s is having a great week or one of those weeks where everything takes longer than it should.

Renata runs a small global health consulting LLC. She works with NGOs and government health programs, which means her client communication is high-stakes and needs to be precise. Half her day used to disappear into email. She was answering questions she'd answered before, following up on proposals, summarizing calls into action items that somehow took 40 minutes to write. She started using AI by building a library of AI-drafted responses for her ten most common client scenarios. She's not sending form letters…she reads every draft to tailor them. But the difference between starting from nothing and starting from something good is, for her, about 90 minutes a day.

What it should NOT look like

There's a version of AI integration that treats every process as something to hand off, every relationship as something to automate, every human judgment as something a model could probably handle. That version is a mistake, and not just philosophically.

The call where a longtime client tells you something's gone wrong? That conversation requires you. The donor who's been with the organization for twelve years and is deciding whether to renew their major gift? That's a relationship, not a process. The difficult staff situation. The moment a customer needs to hear a real person say they understand what went wrong. These aren't inefficiencies to optimize. They're the work.

The line is genuinely not that hard to find. If the task has a formula and produces roughly the same type of output every time and the inputs are consistent, AI can probably help. If it requires your specific history with a person, your judgment in an ambiguous situation, or the kind of trust that only accumulates over time, that stays yours. Always.

Why starting small isn't settling…it's actually the move

There's a temptation, once you see all the places AI could show up in your business, to try to tackle all of them at once. New tools, new workflows, a whole new way of operating. Four months down the road, most of it was abandoned because real business doesn't stop to let you run transformation projects. (Confession: I did that, like a wild mad scientist who was overly excited!)

The businesses that actually get somewhere with AI almost always start with one thing. One task, and one workflow. Something small enough to try in an afternoon and specific enough to tell you quickly whether it's working.

That first win matters more than its size. It gives you a reference point! It tells you “this is what AI actually feels like when it's working.” It builds enough confidence to try the next thing. Which builds confidence for the thing after that. Which is, eventually, a business that runs a little easier than it did before. Maybe not radically transformed, like from a beehive hairdo into a mohawk. Just better in the right places.

That's the accumulation. That's what this is actually about.

One prompt to find your first move

Before you try anything else, run this. It takes five minutes and it'll probably tell you more than an hour of reading about AI will.

 Next issue: the right order to actually do this

Now you have a map of where AI could show up in your operation. The next question, which is the one most people skip, is what order to tackle it in.

Not every good candidate is created equal. Some processes are genuinely ready for AI right now. Others need some groundwork first — not much, but skipping it is why people get inconsistent results and give up. And some should wait entirely.

Stay tuned for our next issue: how to tell the difference, and what week one actually looks like for a real business.