Choosing your model: AI agency services or a branded micro-SaaS

One agent, four salons. Picture this. You built an AI agent that answers booking messages for your cousin's hair salon. It replies to customers, checks the calendar and confirms appointments, even at midnight. A week later, three other salon owners ask if they can have the same agent too. So what do you do next? Do you build a custom version for each of them, or turn it into a product they sign up for and pay for every month? That one choice shapes your prices, your workload, your risk and even your weekends. In this lesson we'll make that choice on purpose, with clear reasons, instead of drifting into it by accident.

What you will learn. Here's the plan. First, you'll learn the two business models. An agency is you, selling your building skills as a service to each client. A micro-SaaS is a small product. SaaS stands for software as a service, which means software people rent through a browser and pay for monthly, like Netflix for business tools. Micro means it's tiny, built by one or two people for one narrow group. Second, we'll put numbers on both so you can see how the money behaves over a year. Third, you'll get four simple questions that tell you where to start.

The AI agency model. In the agency model, you're the builder who does the work for each client. Say a dental clinic wants help qualifying leads. Your agent reads every enquiry from their website, asks about urgency and insurance, and books the promising ones straight into the calendar. You build it inside the clinic's own accounts. You charge a setup fee, which is a one-time payment for the build, and then a retainer, which is a fixed monthly fee for keeping it running and improving it. Think of the retainer like a gym membership for their automation. Money arrives quickly, but every new client needs more of your hours.

The branded micro-SaaS model. Now the other road. Imagine a product called InvoiceNudge. It connects to a small online shop's invoicing tool, sends polite reminders for unpaid invoices, and emails the owner a weekly report of who paid. You build it once. Every customer gets the same workflow, and they only change settings, like the tone of the reminder or how many days to wait. They sign up on your website and pay a subscription, a monthly charge that renews by itself. The catch is that you need time to build it and time to find buyers. But once it works, customer number fifty costs you only a little extra to serve, mostly AI usage and hosting, and no new build.

Side by side. Let's put the two on a scale. On one side sits the toolbox, your agency. It earns early, because a business with a painful problem will pay you this month. But each new client means calls, builds and fixes, so your calendar fills up fast. In return, it can bend to any request, which clients love. On the other side sits the app, your micro-SaaS. It earns slowly at first, but a new customer needs little new work from you. It stays standard, though, and it can't do everything for everyone. Its risk is different too. An agency worries about losing one big client. A micro-SaaS worries that nobody signs up at all.

Worked example: twelve months of income. Let's test this with simple numbers, starting with the agency. It signs one new client a month. Each client pays fifteen hundred dollars setup and five hundred a month. So month one is one client plus a setup fee: two thousand dollars. Every month adds one more client and one more setup fee, so months one to five bring two thousand, twenty-five hundred, three thousand, thirty-five hundred and four thousand. That's fifteen thousand in total. Then your time caps you at five clients. In months six to twelve, five clients pay five hundred each, which is twenty-five hundred a month. Seven capped months make seventeen thousand five hundred. Add the two parts, and the agency's year comes to thirty-two thousand five hundred dollars.

Worked example: the product's year. Now the product. It takes three months to build, then gains eight subscribers a month from month four. Months four to twelve are nine months, so eight times nine gives seventy-two subscribers in month twelve. Each pays forty-nine dollars, so month twelve income is three thousand five hundred and twenty-eight dollars. For the whole year, we add up the subscriber months: eight, then sixteen, and so on up to seventy-two, which makes three hundred and sixty. Times forty-nine, the product's year is seventeen thousand six hundred and forty dollars. So over the year the agency earns almost double. But in month twelve, the product already beats the agency's twenty-five hundred, and it's still climbing. One honest note: these are sales before costs, and they assume nobody cancels. Real numbers will be lower for both models, especially the subscription, once you pay for AI usage, hosting and payment fees, and some customers leave.

Agency first, product later. You don't have to pick one road forever. Here's the shortcut many builders use: let agency clients pay you to do your product research. You build for one client. Then you notice the same request arriving three times, say booking agents for salons. That repeat is your signal. You turn the tested workflow into a template, which is a reusable version with the client details swapped for settings. Add a sign-up page and billing, and you've got a branded micro-SaaS. You start with income from day one and end with a product that real customers already proved they want.

Same engine, different wrapper. Good news for your skills: both models use the same tools. Take n8n, a no-code automation tool where you drag boxes called nodes onto a canvas and connect them, like plugging appliances into a power strip. In an agency, you'd usually set up the workflow inside each client's own account, so they own it and their data stays with them. In a micro-SaaS, one workflow runs in your account and serves every customer, with a login page and a settings form in front. The building blocks are identical. What changes is who owns the workflow, who pays, and how it's packaged.

Four questions to decide. Let's run the salon case through four questions. First, do you need income this month? If you do, lean toward the agency, because setup fees pay quickly. Second, are many businesses asking for the same thing? Our first salon made the request, and three more repeats are pinned on the board, so there's a real product signal. Third, do you enjoy client calls and custom tweaks? That suits agency life. Fourth, are you ready to market to strangers who've never met you? That's what a product demands. For the salons, the answer is clear: start as an agency, but build it like a future template.

Common mistake: product before proof. The most common mistake is building the product before anyone has paid. It feels productive. You spend three months polishing an app with a beautiful dashboard and ten features, launch it, and hear silence. Without a paying customer, you have no proof that the problem hurts enough for someone to spend money on it. Here's the fix. Sell three setups by hand first, even if it's messy and you configure everything yourself. Then look at what all three clients actually used, and turn only that part into your product. Everything else can wait.

Recap. Let's wrap up. An agency sells your building skills: a custom workflow for each client, paid with a setup fee and a monthly retainer. It brings money fast, but your hours set the ceiling. A micro-SaaS sells one product to many customers through a subscription. It starts slowly, then grows without a new build for every customer. And you don't have to choose forever. The smart path for most of us is to start with agency work, watch for requests that repeat, and turn the proven workflow into a product. Same tools, same skills, different packaging.

Your exercise. Before the next lesson, grab a notebook and make your own decision. Pick one workflow you can already build or are close to building, for example an agent that answers customer messages for an online shop, or one that sends invoice reminders. Write your answers to the four questions. Then choose your model: agency, micro-SaaS, or agency first with a product later, and add one sentence on why. Finally, list three real businesses you could offer it to. Keep it to one page. In the next lesson, you'll turn that choice into an offer people actually want to buy.