How to use AI to grow your New York business

Most owners who ask me how to use AI to grow their business are really asking something smaller: where do I even start? There are a hundred tools, every vendor says theirs changes everything, and you have a business to run. Here is how I answer it on a first call.

Start with hours, not tools

Growth usually comes from getting your own time back and putting it on the work only you can do: talking to customers, pricing, hiring, deciding. So skip the tool list and ask a plainer question: which task eats the most hours and needs the least judgment?

For a week, write down every time you or your team do something that feels like moving information from one place to another. Common ones I see:

• Sorting an inbox and writing the same few replies
• Answering the same customer questions by email, text or phone
• Copying details from a form or an email into a spreadsheet or CRM
• Chasing quotes, invoices or follow-ups that slip
• Putting together the same weekly report from several systems

Pick the one that costs the most. That is your first project. Not the whole company, just that one thing.

Know what an agent is good at

An AI agent is software that can read, decide within limits you set, and act in your tools: draft the reply, update the record, book the call. It is good at work that is repetitive, high in volume and has a clear right answer most of the time. It is not good at work that depends on your relationships, your taste or a judgment call with real money behind it.

That split is the whole trick. The agent handles the routine cases. Anything unusual, unclear or high-stakes goes to a person. If an agent is built to guess instead of asking, it will eventually tell a customer something wrong, so I design every one to hand off when it isn’t sure.

Three places New York owners tend to start

1. Your own inbox and calendar

If you are the bottleneck, the quickest win is usually an assistant for you. An AI executive assistant can sort your mail, draft replies in your voice, manage your calendar, write a short brief before a meeting and remind you about follow-ups. You approve anything that goes out under your name. I offer this as the EA for busy professionals, with fixed prices published on that page.

2. Customer questions and new leads

A shop, a studio or a small firm often answers the same ten questions all week. An agent that answers from your approved information, collects the details of a new enquiry and passes the good ones to you means fewer missed leads and fewer evenings in your inbox. You decide what it is allowed to say and what it must pass along.

3. Back-office paperwork

Invoices, intake forms, reports, data entry. This work is dull, easy to get wrong when you are tired, and expensive when a skilled person is doing it. It is often where a custom agent pays for itself, because the hours are easy to count.

Check whether it is worth it before you build

Not every task deserves an agent. Before spending money, work it through on paper:

1. How many hours a week does this task take, and who does it?
2. What is that time worth, counting what the person could be doing instead?
3. What share of cases are routine enough for an agent?
4. What does it cost to build and run, and how long until it pays back?

If the numbers don’t work, the honest answer is not to build it. That is a fine outcome of a conversation, and it is what AI consulting is for: working out where agents would help, where they wouldn’t, and what a sensible first project looks like, whether you build it with me or on your own.

Keep it small, and keep people in charge

The projects that go well are narrow. One workflow, clear limits on what the agent may do, a log of what it did, and a person who looks at the exceptions. Once the first one works and you trust it, add the next. Projects that try to automate everything at once tend to stall.

If you already run an AI assistant or agent, watch the bill too. Model costs can creep up for reasons that are easy to fix, such as sending far more text than needed on every call. A token cost review looks at your usage and ranks the changes worth making.

What to do this week

• Write down the three tasks that cost you or your team the most hours.
• Mark which ones follow the same steps each time.
• Pick one and estimate its hours per week.
• Bring that to a short call and we’ll decide together whether an agent fits.

I’m based in New York and work with businesses across the five boroughs. You can read more on the AI agents for New York City businesses page, or see how I build them on the AI agent development page.

If you’d like to talk it through, book a 30-minute discovery call through Finn, the assistant on sharkbook.net. Bring the task you wrote down, and we’ll see whether it’s worth an agent.