A Tale of Two Sides

Two ways to make money in AI right now. They build completely different businesses, and both of them are good.

October 25, 2026

Read Time: 9 minutes

Win of the Week

Sean sent this over after his first two weeks.

Roughly $179,000 of project value and $8,000 a month of recurring, in a fortnight…

There Are Two Businesses Here

There are two completely different ways to make money in AI right now, and almost nobody separates them properly.

They look similar from the outside. Both involve AI. Both involve selling to businesses. Both get described as "an AI business" at a dinner party.

They are not remotely the same thing. Different clients, different sales motions, different delivery, different economics, and most importantly, different lives at the end of it.

I run both. That is not a flex, it is the reason I can tell you what each one actually involves rather than the brochure version.

And before we go any further, I am not building toward a winner here. Both of these are genuinely excellent businesses. They are just excellent in completely different ways, and the mistake almost everyone makes is choosing between them on numbers rather than on fit.

Side One: Consulting

Bigger projects. Higher values. Longer relationships.

You walk into a business, map how it actually works, find what is broken, and then either build the fix or oversee somebody building it. If you are doing implementation as well as assessment, it gets significantly more complex, and the complexity compounds the deeper you go into their systems.

That complexity is the point.

It is hard, and hard filters people out. Which means that when you get good at it, you are not standing in a queue of forty identical providers being compared on price. You are one of very few people in a given market who can walk into a fifty person business, understand how the money actually moves through it, and produce something that works.

It also moves to money fast, because of the size of the problem. A business losing four hundred thousand a year to a broken process does not haggle over a fifteen thousand dollar fee. The maths does it for them.

And the project is only the front end.

What attaches behind it is maintenance, advisory, and whatever you build for them next. The implementation is the headline number, but the monthly consulting fee and the maintenance agreement are the parts still arriving next summer.

Then there is tenure. When a business finds an AI consulting firm it actually trusts, it stops shopping. You know their systems, their people, and which of their processes are held together by one person's memory. Replacing you means starting that education again with a stranger, and nobody wants to do that.

Partnerships sit on top of it. Once implementation firms know you, they bring you deal flow without you asking.

So consulting is large upfront cash, recurring stacked behind it, from clients who stay for years. That is a very good business.

Side Two: Growth Services

The other side is productised, and it is a different animal entirely.

Call it AI growth services. Websites, marketing, the operational infrastructure a local business runs on day to day. Everyone gets more or less the same thing, and the entire game is driving delivery cost as close to zero as you can.

This is the agency offer I run, so rather than keeping it abstract:

• Website

• AI receptionist

• AI chat

• Five star Google review system

• Review reactivations

• Full AI CRM and interface

$497 up front, then $297 to $697 a month depending on the business.

That is the whole offer. It does not change per client, and that is deliberate, because every piece of customisation you allow is margin walking out of the door.

Now the number that actually matters here, and the one I left out of my own thinking for far too long.

The margins are absurd.

Once delivery is genuinely operationalised, the marginal cost of your forty first client is close to nothing. Not low. Nothing. Eighty percent plus is a normal margin on this, and run properly it gets a great deal closer to ninety nine than most people would believe.

If you are doing the selling yourself, your only meaningful cost is ad spend, and if you are selling without ads then your cost line is effectively empty. There is no delivery team scaling alongside revenue, no hours being consumed, no project risk sitting on your balance sheet.

That is a fundamentally different machine to consulting, where your cost of delivery is you, or people you have to hire and train.

And the market for it is expanding rather than contracting.

The arbitrage AI created means businesses that could never previously afford marketing help can now afford it comfortably. The plumber who was never going to pay a four thousand dollar monthly retainer will absolutely pay five hundred. That is not a smaller version of the old agency market. It is a vastly larger one that did not exist three years ago, and almost nobody is serving it yet.

One call closes it. No scoping phase, no proposal document, no three week wait while a committee forms an opinion. The prospect understands what they are buying inside ninety seconds, because it is the same thing you sold the last forty people.

Recurring from day one, at ninety percent margins, into a market that is still filling up. That is also a very good business.

The Sales Motions Are Nothing Alike

This is the practical difference that actually changes your week.

Consulting is a diagnosis sale. You are not selling a thing, you are selling the finding out. The conversation is long, it is consultative, and most of your skill goes into asking better questions than the last person who sat in that chair. The prospect usually does not know what they need. They know something is wrong.

You win it by being the person who understands their business best, which takes time in the room and cannot be rushed.

The agency is a recognition sale. You are not diagnosing anything. You are describing a known package to somebody who already knows they have the problem. The plumber knows his website is bad. He knows he misses calls. He is not confused about his situation, he just has not got round to fixing it.

You win that one on clarity and speed, not insight.

So the two require genuinely different things from you. Consulting rewards depth, patience and judgement. The agency rewards volume, consistency, and a process you can run four hundred times without getting bored.

Be honest about which of those you are good at, because wanting to be good at something is not the same as being good at it.

Both Roads End at Recurring Revenue

Here is the comparison people get wrong, including me until recently.

It is not that one of these has recurring revenue and the other does not. They both do. The difference is how many relationships it takes to get there.

Eight thousand a month of recurring on the consulting side can come from two or three clients. Maintenance agreements and advisory retainers are big line items, so it does not take many.

The same eight thousand on the agency side is sixteen clients at five hundred. Sixteen onboardings, sixteen support relationships, sixteen separate reasons someone might churn in month nine.

Fewer, deeper, larger. Or more, shallower, smaller.

Neither of those is better. They are different operational problems, and which one you find tolerable says more about you than about the models.

What is worth noticing is that the agency version holds its margin as it scales and the consulting version does not, because at some point consulting growth means hiring consultants. The agency version keeps most of what it bills. The consulting version buys bigger deals and then pays people to deliver them.

Both get you somewhere serious. They just do it with completely different shapes of profit and loss.

What Each One Costs You

The version people sell you is always the good half with the bad half removed, so here is the other half of both.

Consulting is lumpy early and complicated forever. Before the backend stacks up, income arrives in chunks with gaps between them, and the gaps do not care about your rent. Three deals slipping into next month is a cashflow event rather than a rounding error.

The lumpiness resolves once you have maintenance and advisory running underneath you. The complexity does not, and is not supposed to. The work is genuinely difficult, delivery risk sits with you rather than with a platform, and you have to keep getting better because the problems keep getting bigger. If that reads as a cost rather than the interesting part, pay attention to that reaction.

And growth means people. Scaling consulting means finding and training people who can do what you do, which is the hardest hiring problem in professional services.

The agency takes a long time to feel like anything. Three clients at $500 is $1,500 a month, which is not a business, it is an irritation with a Stripe account attached.

You need volume, volume needs a selling machine, and building a selling machine takes months before the numbers look like anything you would tell anyone about. In month four you will be working extremely hard for an embarrassing figure, and that is where almost everybody quits.

The margins are only extraordinary once delivery is genuinely operationalised. Before that, you are doing custom work at productised prices, which is the worst seat in the house. Get the systems wrong and twenty clients will bury you faster than two consulting clients ever could.

And churn is quiet. It undoes months of selling without announcing itself, and you tend to notice three months after it started mattering.

Both of those are real. Anyone telling you otherwise is selling you something.

Can You Run Both?

Eventually. Not from a standing start.

They demand different behaviours on the same Tuesday. Consulting wants you deep inside one business for a day. The agency wants you on the phone forty times. Split yourself across both before either is working and you will do both badly.

If you do want both, there is an argument for consulting first, because it reaches cash quickest and it teaches you what local businesses actually need, which is the only reliable way to design a productised offer people want. Every productised thing I have built came out of doing the bespoke version enough times to see the pattern.

But that is a sequence, not a hierarchy. Plenty of people should start on the agency side and never touch consulting, and they will do perfectly well.

Pick one. Get it running without your full attention. Then add the other if you still want it.

Which One Fits You

Here is where I have landed, and it is as much self-reflection as advice.

Start with who you want to sit across from.

Some people are energised by a room with a fifty person manufacturer's operations director, working out why their quoting process loses two weeks a month. Other people would far rather sell to the plumber, the HVAC guy, the roofer, the salon owner, the barber. Straightforward people, straightforward problems, something you can hand over on Friday.

Neither is the better person to be. But one of them is you, and you probably knew which halfway through that paragraph.

Then the shape follows.

If you want to close a lot of people, take smaller amounts, and build something highly predictable at very high margin, the agency side is significantly easier to build a brilliant lifestyle business around, and it scales aggressively once the machine runs. The market for it is bigger than it has ever been.

If you want fewer, larger relationships, work that keeps getting harder in an interesting way, and a business built on difficulty being a barrier, consulting is the one. It reaches serious money quickly and the relationships last years.

Neither of these is the smart choice. They are answers to two different questions, and the question was never which model makes more money. Both make a great deal of money when they are run properly.

The question is what you want your week to look like in three years, and who you want to spend it with.

The Real Point

One side is large projects with maintenance and advisory behind them, clients who stay for years, and a moat built out of difficulty.

The other is one offer sold hundreds of times, closed on a single call, at margins the consulting side cannot touch, into a market AI has only just made affordable.

Both end at recurring revenue. One gets there with three relationships, the other with sixteen.

So work out who you actually want to spend your Tuesdays with, and pick that one deliberately rather than drifting into whichever you heard about first.

See you next week,

– Andrew

P.S. If you want to work with me via coaching like Sean, reply back to this email.

💡 Two ways to start

Launch a local AI agency. We build it for you in 24 hours. Brand, website, offer, demo assets, demo script, proposal and contract, all in your name. You sell the 24/7 AI receptionist & 5-star Google Reviews, and our platform handles delivery.

Launch a local AI agency

Launch an AI consultancy. Use your existing industry expertise to run paid AI assessments for businesses. Audits from $5,000 to $50,000, leading into implementation and ongoing retainers.

Launch an AI consultancy