Thorsten Dembski | AI Agency Side Hustle

Start an AI agency on the side: sell AI agents, automations & websites to local businesses. Free step-by-step guide ↓

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

If a plumber can't repeat your offer after hearing it once, you don't have an offer yet.

Most people trying to sell AI services say something like:

"I build AI-powered automation solutions for businesses."

That sentence could mean anything. So it means nothing.

Use this instead:

"I help [specific people] get [result they can picture] without [pain they hate]."

Examples (made up, to show the shape):

"I help dental practices stop losing patients to unanswered calls, without hiring anyone."

"I help estate agents follow up with every enquiry within minutes, without chasing."

"I help lawyers book consultations around the clock, without extra staff."

3 rules:

Name the person. "Businesses" is nobody.

Name a result they can picture. "Efficiency" isn't a picture. "No missed calls" is.

Use words your mother would use. AI is fine as the how. It's never the what.

Now your turn.

Write your offer in one sentence in the comments 👇

I'll reply to the first 20 with a rewrite.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

A close friend of mine was scared.

Skeptical. Hesitant. Every reason in the world not to do it.

I pushed him anyway.

He bought 10 apartments.

Later he sold 8 of them for a 500% gain. The 2 he still owns, he effectively got for free.

Every time we meet, he thanks me for pushing him.

(One friend. One market. One time. His result, not a promise.)

I didn't push him because I wanted my commission.

I pushed him because I knew it was right for him, and I could see that the only thing standing in his way was fear.

This is the most important lesson of my career:

Most "no" is not a no.

It's fear.

And the same thing will happen the day you try to sell an AI agent, a website or an automation to a local business owner.

He needs it. He wants it. He can afford it.

And he says, "I'll think about it."

Here are the 5 fears behind that sentence, and what moves each one:

"I'll waste my money."

Make the first step small. A short pilot instead of a long contract.

Make the cost of waiting visible: "What does it cost you each month this stays as it is?" Let him say the number.

"I'll look stupid."

Give him something he can show his partner or his team. Proof. A result on one real case. A one-page summary he can forward.

"I'll be locked in."

Month to month. Easy exit. "If it isn't working, you stop. No hard feelings."

Removing the exit is what scares people. Opening it is what lets them walk in.

"It won't work for MY business."

Show it on his business before he buys. His name. His calls. His website. Generic demos don't move him. His own does.

"I have to decide alone."

Ask: "Who else should see this?" Invite the office manager, the partner, the spouse into the next call.

Then the push.

A push is a gift when 4 things are true:

He truly needs it.
He truly wants it.
He can afford it.
You'd bet your own money on it.

If any one of these is missing, you don't push. You let him go.

That's the difference between selling and hustling.

The test I use is simple: "Will he thank me in a year?"

If yes, you owe him the push.

Because the real price of a "no" born from fear isn't paid by you.

He pays it, every month, for years.

Save this for the day a perfect-fit prospect says "let me think about it."

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

Think of the most expensive thing you ever bought on a gut feeling.

A car. A course. A holiday. A house.

Now think about the moment you said yes.

I've watched people buy for 30 years. It almost always comes down to the same 4 things:

They trusted the person.
They could picture themselves with the result.
Staying where they were started to hurt more than moving.
Someone actually asked.

Logic doesn't decide. Logic justifies.

First comes the feeling. Then the brain builds the reasons so it feels safe.

That's why a long feature list loses and a clear picture of "your Saturdays back" wins.

So tell me:

What did you buy on a gut feeling?

What was the exact moment you said yes?

Drop it below 👇

I'll reply to as many as I can with what it tells you about how to sell.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

If a plumber can't repeat your offer after hearing it once, you don't have an offer yet.

Most people trying to sell AI services say something like:

"I build AI-powered automation solutions for businesses."

That sentence could mean anything. So it means nothing.

Use this instead:

"I help [specific people] get [result they can picture] without [pain they hate]."

Examples (made up, to show the shape):

"I help dental practices stop losing patients to unanswered calls, without hiring anyone."

"I help estate agents follow up with every enquiry within minutes, without chasing."

"I help lawyers book consultations around the clock, without extra staff."

3 rules:

Name the person. "Businesses" is nobody.

Name a result they can picture. "Efficiency" isn't a picture. "No missed calls" is.

Use words your mother would use. AI is fine as the how. It's never the what.

Now your turn.

Write your offer in one sentence in the comments 👇

I'll reply to the first 20 with a rewrite.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

You didn't lower your price because it was too high.

You lowered it because you couldn't tell a buyer from a browser.

Codie's two reasons are one loop.

A browser asks "how much?" in the first minute. You say the number. He says "bit much." You think it's the price, so you drop it.

Now you have ten more browsers and no time for the one buyer.

Cheap price, wrong customer. Same mistake, run twice.

On the phone you find out in five minutes. There is no brochure to hide behind.

Here is what I ask, before I ever say a number:

"What made you reach out this week and not last month?"
"What have you already tried?"
"What is it costing you every month?"
"On a scale of 1 to 10, how much do you want it gone?"

The buyer has a reason, has tried something, knows his number, and says 8 or more.

The browser has none of that. He says "sounds great, send me the details."

Then the hard part: don't fix his problem while he's still talking. Let him hear himself say the number.

When all four land, the price is not an argument. He already told you what it's worth.

Nobody taught you this, so you guessed, and the guess cost you money.

Charging more is a selling skill, not a pricing trick.

Try "why now?" on your next call. Watch who goes quiet.

[Codie Sanchez @Codie_Sanchez]

Only 2 reasons 99% businesses never scale:

1. They charge too little
2. They sell to the wrong customer

Aka they chase volume so end up serving more people but worse.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

A roofer paying $228 a lead for a $32,000 roof.

From the video, here is what that means when he doesn’t sell the funnel but the phone.

Say 40 leads a month. If the roofer books 1 job from every 6, that is almost 7 roofs.

If the phone gets that to 1 in 5, it is 8 roofs.

Same ads. Same funnel. One better conversation out of every 30 leads.

That is roughly $40,000 more a month, and it did not cost him a single extra lead.

(Example numbers taken from his video, the math is mine.)

He says it himself near the end: when leads go up and jobs don’t, the problem is the phone.

Then the video moves on, because the phone is not a funnel.

But it is the place where a $228 lead becomes a roof or becomes a missed call.

How many of those 40 leads get a real conversation in the first five minutes?

Build the funnel and ignore the phone, and you built the cheap half.

[The Boring Marketer @boringmarketer]

HOW TO BUILD YOUR AI SERVICES BUSINESS (targeting Main Street "boring businesses")

In 39 minutes:
- why you shouldn't sell ai automation, seo/aeo, or ad services
- why you shouldn't go after HVAC and plumbing companies
- why you should sell the ai reinvention of a $15,000-$20,000 service for $3,000-$5,000 with SOFTWARE margins

and exactly how I would do it...with a full funnel for a roofing company included

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

A close friend of mine was scared.

Skeptical. Hesitant. Every reason in the world not to do it.

I pushed him anyway.

He bought 10 apartments.

Later he sold 8 of them for a 500% gain. The 2 he still owns, he effectively got for free.

Every time we meet, he thanks me for pushing him.

(One friend. One market. One time. His result, not a promise.)

I didn't push him because I wanted my commission.

I pushed him because I knew it was right for him, and I could see that the only thing standing in his way was fear.

This is the most important lesson of my career:

Most "no" is not a no.

It's fear.

And the same thing will happen the day you try to sell an AI agent, a website or an automation to a local business owner.

He needs it. He wants it. He can afford it.

And he says, "I'll think about it."

Here are the 5 fears behind that sentence, and what moves each one:

"I'll waste my money."

Make the first step small. A short pilot instead of a long contract.

Make the cost of waiting visible: "What does it cost you each month this stays as it is?" Let him say the number.

"I'll look stupid."

Give him something he can show his partner or his team. Proof. A result on one real case. A one-page summary he can forward.

"I'll be locked in."

Month to month. Easy exit. "If it isn't working, you stop. No hard feelings."

Removing the exit is what scares people. Opening it is what lets them walk in.

"It won't work for MY business."

Show it on his business before he buys. His name. His calls. His website. Generic demos don't move him. His own does.

"I have to decide alone."

Ask: "Who else should see this?" Invite the office manager, the partner, the spouse into the next call.

Then the push.

A push is a gift when 4 things are true:

He truly needs it.
He truly wants it.
He can afford it.
You'd bet your own money on it.

If any one of these is missing, you don't push. You let him go.

That's the difference between selling and hustling.

The test I use is simple: "Will he thank me in a year?"

If yes, you owe him the push.

Because the real price of a "no" born from fear isn't paid by you.

He pays it, every month, for years.

Save this for the day a perfect-fit prospect says "let me think about it."

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

Think of the most expensive thing you ever bought on a gut feeling.

A car. A course. A holiday. A house.

Now think about the moment you said yes.

I've watched people buy for 30 years. It almost always comes down to the same 4 things:

They trusted the person.
They could picture themselves with the result.
Staying where they were started to hurt more than moving.
Someone actually asked.

Logic doesn't decide. Logic justifies.

First comes the feeling. Then the brain builds the reasons so it feels safe.

That's why a long feature list loses and a clear picture of "your Saturdays back" wins.

So tell me:

What did you buy on a gut feeling?

What was the exact moment you said yes?

Drop it below 👇

I'll reply to as many as I can with what it tells you about how to sell.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

I was very successful in insurance.

Then my bosses wanted the team to sell more of certain contracts.

I thought they were bad. They cost people money, and gave them no real value.

My sense of truth and justice rebelled.

I could not sell something I knew was worse than the alternatives.

So I quit.

Then I moved into real estate: luxury apartments and buildings for wealthy clients who got large tax benefits that made them richer.

Same skills. Different product.

And one thing changed completely: I could sell it with my whole chest.

Here's why that matters to you, if you're about to sell an AI agent, a website or an automation to a local business:

The buyer feels your certainty before they hear your words.

You need to be sure about 3 things:

The product. Does it really do what you say?
Yourself. Are you someone who delivers?
Your company. Even if it's just you, can they trust that you'll be there next month?

If one of them is weak, the buyer senses it in seconds. Then they say, "Let me think about it."

How to build certainty before the call:

Use it yourself first.

Build it for your own business or a friend's. Watch it work. Watch where it fails. Fix that.

Collect real facts and real stories.

One true sentence of what it did for one real business beats ten features. If you have none yet, your own test is the story. Only tell what's true.

Know the alternatives.

If something else is a better fit for this person, say so. You'll lose one sale and win the trust of everyone who hears about it.

Ask: "Would I buy this, from me, at this price?"

If the answer is no, fix the offer. Not the pitch.

The best salespeople I know aren't the best talkers.

They're the most certain.

And certainty is built before the call.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

You don't need a big idea.

You need one of these 7.

Each one is built with AI. Each one has a customer who already has the problem.

The AI receptionist

You build: an agent that answers the phone, books appointments and hands urgent calls to a human. Who pays: dentists, vets, salons, plumbers. Anyone whose phone rings when their hands are busy. First customer: call 5 local businesses after 6pm. See who picks up.

The missed-call text-back

You build: a call goes unanswered, a text goes out within seconds. Who pays: any business that loses jobs to voicemail. Why it's easy: you can explain it to a plumber in one sentence.

The website that books appointments

You build: a clean site with a booking calendar and an enquiry form that lands in a customer list. Who pays: businesses with no site, an outdated one, or one that never sends an enquiry. First customer: Google Maps, search a trade, open the sites. You'll find them in 10 minutes.

The follow-up machine

You build: quotes and old enquiries get automatic reminders until the person answers. Who pays: roofers, cleaners, garages, anyone who sends quotes and then waits.

The review machine

You build: every happy customer gets a message asking for a review. Replies are drafted for the owner. Who pays: any local business that lives on its Google rating.

The content service

You build: 30 days of posts for a local business, written and designed with AI, scheduled automatically. Who pays: the owner who knows he should post and never does.

Your own channel or product

You build: an Instagram, TikTok or Pinterest channel, an AI character, an ebook, a newsletter. Who pays: the audience, the brands, or both. This is the one where you are the business.

What they all share:

One problem. One person who pays. One conversation to start.

And you can run them side by side. Each client gets their own set-up, their own customer list, their own automations. It stays organised when client number 2, 3 and 10 arrive.

Building each of these takes an afternoon, not a year.

The hard part isn't picking the best one.

Pick the one you could explain to a plumber in one sentence.

Which number? 👇

I'll reply to the first 20 with the first person I'd talk to.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

You're not losing sales because of your product.

You're losing them because of one of these 5 things.

A. Nobody can tell what I sell.

Fix: One sentence. "I help [who] get [result] without [pain]." Test it: would a plumber understand it?

B. I never really ask for the order.

Fix: Ask on the first call. Then stop talking. The silence after the ask belongs to them.

C. I pitch before I ask.

Fix: Two questions about their business before one word about yours.

D. They say "I need to think about it" and I go quiet.

Fix: It's a doubt, not a no. Ask what exactly they want to think about.

E. I haven't talked to enough people yet.

Fix: Count your real sales conversations this week. If the number is 0 to 3, that's your problem. Not your product.

Be honest. Pick the one that hurts.

Comment the letter 👇

I'll reply to every one with what to fix first.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

I was very successful in insurance.

Then my bosses wanted the team to sell more of certain contracts.

I thought they were bad. They cost people money, and gave them no real value.

My sense of truth and justice rebelled.

I could not sell something I knew was worse than the alternatives.

So I quit.

Then I moved into real estate: luxury apartments and buildings for wealthy clients who got large tax benefits that made them richer.

Same skills. Different product.

And one thing changed completely: I could sell it with my whole chest.

Here's why that matters to you, if you're about to sell an AI agent, a website or an automation to a local business:

The buyer feels your certainty before they hear your words.

You need to be sure about 3 things:

The product. Does it really do what you say?
Yourself. Are you someone who delivers?
Your company. Even if it's just you, can they trust that you'll be there next month?

If one of them is weak, the buyer senses it in seconds. Then they say, "Let me think about it."

How to build certainty before the call:

Use it yourself first.

Build it for your own business or a friend's. Watch it work. Watch where it fails. Fix that.

Collect real facts and real stories.

One true sentence of what it did for one real business beats ten features. If you have none yet, your own test is the story. Only tell what's true.

Know the alternatives.

If something else is a better fit for this person, say so. You'll lose one sale and win the trust of everyone who hears about it.

Ask: "Would I buy this, from me, at this price?"

If the answer is no, fix the offer. Not the pitch.

The best salespeople I know aren't the best talkers.

They're the most certain.

And certainty is built before the call.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

You don't need a big idea.

You need one of these 7.

Each one is built with AI. Each one has a customer who already has the problem.

The AI receptionist

You build: an agent that answers the phone, books appointments and hands urgent calls to a human. Who pays: dentists, vets, salons, plumbers. Anyone whose phone rings when their hands are busy. First customer: call 5 local businesses after 6pm. See who picks up.

The missed-call text-back

You build: a call goes unanswered, a text goes out within seconds. Who pays: any business that loses jobs to voicemail. Why it's easy: you can explain it to a plumber in one sentence.

The website that books appointments

You build: a clean site with a booking calendar and an enquiry form that lands in a customer list. Who pays: businesses with no site, an outdated one, or one that never sends an enquiry. First customer: Google Maps, search a trade, open the sites. You'll find them in 10 minutes.

The follow-up machine

You build: quotes and old enquiries get automatic reminders until the person answers. Who pays: roofers, cleaners, garages, anyone who sends quotes and then waits.

The review machine

You build: every happy customer gets a message asking for a review. Replies are drafted for the owner. Who pays: any local business that lives on its Google rating.

The content service

You build: 30 days of posts for a local business, written and designed with AI, scheduled automatically. Who pays: the owner who knows he should post and never does.

Your own channel or product

You build: an Instagram, TikTok or Pinterest channel, an AI character, an ebook, a newsletter. Who pays: the audience, the brands, or both. This is the one where you are the business.

What they all share:

One problem. One person who pays. One conversation to start.

And you can run them side by side. Each client gets their own set-up, their own customer list, their own automations. It stays organised when client number 2, 3 and 10 arrive.

Building each of these takes an afternoon, not a year.

The hard part isn't picking the best one.

Pick the one you could explain to a plumber in one sentence.

Which number? 👇

I'll reply to the first 20 with the first person I'd talk to.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

You're not losing sales because of your product.

You're losing them because of one of these 5 things.

A. Nobody can tell what I sell.

Fix: One sentence. "I help [who] get [result] without [pain]." Test it: would a plumber understand it?

B. I never really ask for the order.

Fix: Ask on the first call. Then stop talking. The silence after the ask belongs to them.

C. I pitch before I ask.

Fix: Two questions about their business before one word about yours.

D. They say "I need to think about it" and I go quiet.

Fix: It's a doubt, not a no. Ask what exactly they want to think about.

E. I haven't talked to enough people yet.

Fix: Count your real sales conversations this week. If the number is 0 to 3, that's your problem. Not your product.

Be honest. Pick the one that hurts.

Comment the letter 👇

I'll reply to every one with what to fix first.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

I was the best in my company on the phone.

Best at getting leads. Best at turning leads into appointments. Best at selling directly over the phone.

10 years later I watched The Wolf of Wall Street.

In the phone scenes, I recognised myself.

Not the chaos. The calls.

Enthusiastic. Direct. Professional. Kind, and sharp at the same time. Voice going up and down. Almost overacting.

I had done all of it by feel. I never had words for it.

Here are the words.

The first 10 seconds of any call must do 5 things:

Sound sharp.

Short sentences. No "um". No "sorry to bother you".

Every apology tells the other person you're a problem.

Sound glad to be there.

Energy is contagious. The person on the other end copies yours.

Flat voice in, flat voice out.

Sound certain.

Let your voice go down at the end of a sentence. A voice that goes up sounds like asking permission. A voice that goes down sounds like a decision.

Move your voice.

Monotone is the fastest way to get hung up on. Go up, go down, speed up, slow down. Almost overact. On the phone, only the exaggerated version arrives as normal.

Say who you are and why the next 30 seconds are worth it. In one breath.

Example:

"Hi Maria, it's Thorsten. Quick one. 30 seconds, then you decide if it's worth more."

Short. Warm. Certain. And it gives her control, which is why she stays.

Try this today:

Record yourself saying your opening line. Listen to it.

Would you buy from that person?

Most people cringe. Good. That's the work.

Local business owners still answer their phones.

You can build an AI agent in an afternoon. It will never matter if the first 10 seconds of your call lose the owner.

Save this. Record your opening line before you sleep.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

My first day in sales, I knew nothing.

No experience. No training. No idea this would become my life's work.

I'd started at an insurance company. Someone handed me a book called "Become an expert salesman in 7 days".

I did not become an expert in 7 days.

But I was hooked.

Because the book was hardly about tactics. It was about mindset. How a salesperson thinks before they ever say a word.

That was about 30 years ago.

I've read hundreds of books since. Sat in countless seminars and workshops. And I still learn something new every day.

Here's what that first day planted in me, and what I'd tell anyone starting with zero experience today:

You don't need experience. You need a first conversation.

Nobody is good at the start. The only people who got good are the ones who had a first day.

Mindset comes before technique.

If you believe selling means bothering people, your voice says it in the first 4 seconds. No script can fix that.

Selling is helping someone make a decision that's good for them. Believe that first.

Selling is a skill. Not a gift.

You learn it like a language. Badly at first. Then fluently.

Every conversation is a rep.

Books gave me the spark. Calls gave me the skill. You need both. Most people stop at the book.

Be careful who teaches you.

Ask one question: "What have you sold?" If there's no real answer, keep your money.

Now look at where we are.

AI lets you build a website, an automation or an agent that answers the phone for a dentist in an afternoon.

That part is solved.

The part that decides whether you earn money is the part that took me 30 years and still isn't finished: sitting across from a real person and helping them say yes.

You're not late.

I started knowing nothing, and the only thing that changed it was a first day.

Make this week your first day.

What's yours going to be? 👇

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

Perfect tool. Perfect font. Perfect color. Perfect niche.

None of it has ever made a dime.

Pick one thing to build. Build it in a day. Ugly if you have to. Put it in front of one real person. Ask for the money.

One "no" teaches you more than 30 days of research.

Speed beats perfection. Every time.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

@rishabhnasa Thanks for your kind words. Appreciate the feedback. And yes, you absolutely need to have a great product and you need to be convinced. Then focus on the sales process and how it helps your prospects.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

3 calls. 2 businesses said yes on the call. The 3rd signed a week later.

If you can’t sell what you build, you have a hobby. Not a business.

Here’s the story, and the 5 things my script was built on 👇

A week earlier I’d spent a whole week building an AI agent. It worked. I was proud of it.

Then I pitched it to 5 businesses. 0 yes.

Same product. Same service. Same price.

So I stopped tweaking the product. I went back to everything I’d learned in years of selling, wrote a script, and at 4pm I called 3 businesses.

2 said yes on the call.

Here’s what the script was built on:

1. The first 10 seconds decide everything.
People decide if you’re sharp, excited and know your stuff before they hear one feature. So: one specific reason for calling. Energy. No life story.

2. Ask before you pitch.
2-3 questions about their situation and what it’s costing them. Their answers ARE the pitch.

3. Show 2-3 benefits. Not 12 features.
Nobody buys a feature list. They buy what changes for them.

4. Ask for the order. Then shut up.
Clear, direct, no trailing off. The first one who talks loses.

5. “I need to think about it” is a doubt, not a no.
Behind every doubt is one of three things: “I’m not sure it works.” “I’m not sure about you.” “I’m not sure about your company.” Find which one, answer that one, ask again. Plan for 3 rounds, not 1.

AI lets you build a store, a channel, a course or an agent in a week now.

That part is solved.

The money sits on the other side of a conversation.

Build fast. Sell more.

Save this for your next pitch. And follow me: every day I post how to build with AI and how to sell it.

[Thorsten Dembski | AI Agency Side Hustle @ThorstenDembski]

Picture two guys. Same idea: an AI influencer on TikTok.

Peter researches the perfect niche. Compares every tool. Can't decide. Plans for months.

When he finally builds it, he has no idea what to post or how to sell anything with it. No views. No motivation.

"This AI stuff doesn't work."

Paul picks a man or a woman. Generates 50 faces in one go. Keeps the best one. Makes a character sheet so it looks the same everywhere. Picks a product in a proven niche - health, wealth, money, fitness. Starts posting.

A few weeks later, character #2 is in production.

Same tools. Same internet.

Paul started messy and learned to sell while he built.

Be Paul.