That competition is good for Main Street. But it’s a race to save owners time, and time isn’t what keeps them up at night. What keeps them up is where the next customer will come from.

A word on where I sit. I run Alignable, a network of 10 million small business owners, so I have a stake in this argument. Alignable also co-hosts small business AI training with Anthropic, one of the companies named here. Read what follows with that in mind. But I also have something the assistants don’t: 13 years of data on how small businesses actually find work.

Fifty beats 500

Start with where customers come from. In our surveys, members say about 85% of their customers come through referrals. Not ads, not search, not cold email. People who already trust them.

Across 35,000 local communities, our members have formed 226 million business connections. Most of them never become a relationship that sends work either way. The owners who grow through their networks aren’t the ones with the most connections. I’ve told owners for years that 50 warm relationships beat 500 cold contacts. That’s a rule of thumb from experience, not a number we’ve measured. But the owners who win do keep a small set of relationships active and leave the rest alone.

And most owners are sitting on more opportunity than they know. When members tell us what their ideal customer looks like, we can count how many of those businesses sit one introduction away: connected to someone the owner already knows. Among members whose ideal customers show up in their extended network, the typical one has about 190 of them a single introduction away. For one in four, it’s more than 600.

The next customer is rarely a stranger. They’re one relationship away, through someone the owner hasn’t talked to in months.

Those relationships aren’t random, either. In my experience, three kinds do most of the work: customers who talk you up, businesses that serve the same customer without competing with you, and a handful of people who open doors. The best referral partner usually isn’t a peer in your own industry. It’s the business your customer calls right before or right after they call you: the realtor and the home inspector, the accountant and the financial planner.

What a general assistant can’t see

Ask any of the big assistants how to get more referrals and you’ll get a solid list of best practices. Ask an assistant that sits on a map of business relationships and it can tell you something specific: three nearby businesses serve your ideal customer, one of them already knows two of your clients, and you haven’t spoken to your best referral partner in four months.

That’s the difference between advice and action. A general assistant can’t answer that question, because the answer lives in data it doesn’t have.

The early numbers bear this out. In Bluehost’s 2026 study of 350 U.S. owners, 78% said AI saves them time every week. But 43% have seen no change in revenue. Time saved only matters if it goes somewhere that pays.

Too often it goes into volume. Forbes recently told the story of Paul Birrell, a New Jersey owner whose automation loop sent roughly 400,000 emails in three days. His reaction: “I don’t have 400,000 customers.” Everything we see says growth comes from the opposite: judgment about the few people who matter.

Three kinds of AI, not one

AI will reach small business owners in three forms. The first is the personal assistant the giants are fighting over. It delivers time, and the assistants are converging fast.

The second is AI built into the systems that already hold the business’s own data. Intuit knows the books. Shopify knows the store. That AI delivers control: it can flag a cash crunch six weeks out or spot the product quietly losing money.

The third is AI built on the relationships between businesses: who knows whom, who serves the same customers, who sends work to whom. That’s the one that delivers growth, and it’s the one almost nobody is talking about. At Alignable, ours is called Allie.

The giants have real advantages, and they’ll connect to the other two kinds. Meta’s Muse already plugs into QuickBooks, Shopify and Stripe, and the others are adding similar connections. The open question is who the intelligence works for. Meta makes nearly all of its money selling ads, and an assistant built by an ad company will tend to find that the answer involves more advertising. The data that actually works for the owner describes how their business runs and who in their community can help it grow.

What owners should do now

Use a personal assistant every day. The time savings are real.

Then decide, in advance, where the reclaimed hours go. Put them on the calendar against the work that generates revenue, and the most neglected item on that list is relationships. Write down the 50 people most likely to send you business. Most owners can name 10 without trying. Those 10 are where AI should be pointing you.

There are about 36 million small businesses in the U.S., and more than 80% of them have no employees. For those owners, AI isn’t a department upgrade. It’s a capability they never had. Owners should take the better tools the giants’ race produces. But the biggest gains won’t come from saving an hour. They’ll come from knowing which 50 relationships to spend it on.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.