AI setter software vs done for you: who should run your DMs?

A fitness coach told me on a call how his AI setter tool lost leads. When a lead’s answer didn’t match the script, the AI missed the detail and asked the same question again, because that question came next. “And then it looks like it’s AI and they normally either call it out or just stop responding.” My answer: “Rookie mistakes. Not something I’ve not seen before.”

Coaches usually come to us after trying the $99-a-month tools. Another coach, who works with chronic back pain, said his tool followed up “way too fast” with people who are in pain and have their guard up. A trading educator who had tried three AI setter providers no longer wanted anything he’d have to micromanage and “keep fixing when it breaks.”

Each of those problems is a job somebody has to do: tuning the script, timing the follow-ups, fixing what breaks. With a tool, those jobs are yours. For 3+ years my team at OmniFusion has done them for 120+ high-ticket coaches.

What’s the difference between AI setter software and done for you?

With software, you buy an engine built for every business out there, and you become the team that runs it. With done for you, you hire the team and the software comes with it. As I put it on one call: “Our product is not a software in itself. It includes software, but this is a service.”

Both take the texting off a human setter. The difference is everything around the texting, and who does it. Here’s how the three options compare for a high-ticket coach whose leads come in through Instagram DMs:

What matters	In-house setters	AI tool you run yourself	Done for you
Reply speed	When someone is on shift	Fast, day and night	About 2 minutes, day and night
Sounds like you	Depends on each hire and their training	As good as the prompts you write	Trained on your DMs, content and sales script
Daily owner	You or a sales manager	You, or someone you hire for it	A team of 10 salespeople and AI developers
Testing scripts	Hard to track across people	Yours to set up and measure	Openers and scripts A/B tested every month
After Meta’s 24-hour window	Your setters, if they remember	Someone you find and train	A person we provide and train
Control	Full: you hire and you decide	Full: you own the setup and change it any time	You approve the tests, we make the changes
What it costs you	$5,000 to $10,000 a month, plus commissions	A subscription, plus the hours of whoever runs it	One fixed cost that stays predictable as you grow
Best fit	Phone setting, and the early days before your DMs are proven	Low volume, a simple offer and a technical owner	Proven DMs and more conversations than your team answers well

The tool wins on control and on the monthly bill, and in-house setters still own the phone and the early days. Done for you wins once your DM process is proven and the volume has outgrown your team, and that’s the only situation we take on.

If you’re still deciding whether AI should take the texting from your setters at all, start with AI setter vs human setter. If you’ve already decided it should, the real question is who runs it.

What would you have to run yourself with an AI setter tool?

Everything around the messages. The tool sends them. Somebody on your side still owns these jobs, every day:

• Training it on your voice. Your past DMs, your content, your sales script and your answers to the questions leads ask most. When we onboard a coach, my team rescans their past couple thousand DMs first to find where leads stop replying and which objections keep coming up.
• Guardrails. A list of what the AI must never say. Money questions belong in the booking form, out of the DMs. One of our clients had account restrictions because of a savings question in their sales script. It had nothing to do with the AI, and the restrictions stopped once the question was gone.
• Pacing that reads human. Random delays, short messages, follow-ups timed to the person. Too fast and it reads like a bot. Too slow and the lead is gone.
• Opener and script tests. Writing variants, splitting leads between them and measuring the reply rate of each message. We had to build our own software to see that number. We run 3 to 5 opener tests a month, and for Kneehab Academy we’ve run more than 50.
• Reading conversations. For the first 60 days with a new client, my team is in the inbox every day, reading conversations and retraining the AI on the weak ones. Someone has to catch the repeated question before the lead does.
• Meta’s 24-hour window. Automated replies run on Instagram’s messaging API inside Meta’s standard messaging window, which lasts 24 hours for most user actions. After it closes, the follow-ups have to come from a person. That’s an hour or two a day, every day.
• Booking and show-up. A booking link per lead, tied to the conversation it came from, then reminders on WhatsApp, SMS or iMessage until the call. Our target is 70 to 80% of booked calls showing up.
• The loop with your closers. Your closers mark every call qualified or not, closed or not, and that has to flow back into the qualification. Without it, the AI keeps booking the same wrong people.
• Hand-offs and the off switch. Leads who ask twice whether it’s AI, partnership requests, friends and family. Someone decides when a person takes over, and someone picks it up when the tool flags it.

Add that up and you’re running an AI department in-house, and managing it every day. You’ve moved the headache from people to a tool, and a badly trained tool can hurt your brand very fast, because it makes the same mistake in every conversation it runs.

They’re the same jobs you have when managing setters and closers: scripts to retrain, conversations to review, numbers to chase. Only now the one you manage is software.

Why is a good AI setter harder to build than it looks?

Because it takes sales skill and engineering together, and most teams have one or the other. You can put 10 salespeople in a room and they won’t build the best AI setter. Put 10 developers in a room and they won’t either.

I’ve been in sales for more than 10 years. Emil, my co-founder, has automated Instagram for business clients for more than 10. We started building this in mid-2023, about six months after ChatGPT came out, and the first year and a half to two years were really hard. The bar wasn’t a VA. It was a trained salesperson in Instagram DMs, and replacing every piece of that person’s intuition took far longer than it looks from the outside.

We also tried a done-with-you version, where the coach’s team ran part of the work. It didn’t work. If we’re responsible for the result, we need control of everything between the first DM and the call. That’s why we only do it done for you.

When does running an AI setter tool yourself make sense?

When someone on your team will own it every day, your volume is low and your offer is simple. In that case a tool can be the right buy:

• You have a technical person whose job it is. Someone who reads conversations every day and fixes the prompts, as their actual job. Not you between client calls, and not a VA who also clears the inbox.
• Your volume is low. With a handful of new conversations a day, one wrong message costs you one lead, and you’ll notice it.
• Your offer is simple. The fewer objections a lead brings, the less the AI has to handle well.
• You want to own the setup and change it yourself. That’s a fair reason. It’s also one of the reasons a coach isn’t a fit for us.

If nothing in your DMs is proven yet, don’t buy a tool and don’t hire us. I always advise doing it yourself first, or hiring an experienced human setter. A person can try ten different things in a day and find out which wording, which script and which kind of lead works for your offer. The AI needs those data points to build on.

And if you personally chat in your DMs and want to keep doing that, keep doing it.

Why hire a team instead of running the tool yourself?

For the same reason you don’t cut your own hair. Anyone can. It’s probably a bad idea. You go to someone who has already done 1,000 haircuts, and in AI appointment setting, that’s what my team is.

Here’s what done for you means with us:

• It’s hands off. My team of 10 salespeople and AI developers runs it as your appointment setting team, so you don’t hire an AI expert or a sales manager, and we stay your setting team however big you get.
• It sounds like you. We built the most human-like AI in the space: trained on your DMs, content, sales script and training material, with reply times that vary like a person’s and messages that mirror each lead’s language and tone.
• You never miss a DM or a follow-up again. It replies in about 2 minutes, day and night, and after Meta’s 24-hour window a person we provide and train takes the follow-ups.
• Nothing goes live untested. You test it on a separate Instagram account first, you choose a staged start or all at once, and one click switches it off for any conversation.
• It keeps getting better. Scripts are A/B tested all the time to lift your booking rate, and we run a closed loop with your closers, because the goal is booked calls with people who pay.
• Changes are quick. You send feedback in our shared WhatsApp or Slack group, and it’s usually changed the same or next day.

The full walkthrough, from the first message to the booked call, is on our done-for-you AI appointment setter page.

What happens when a team runs it?

Jordan Pham runs BNB Success, which teaches Australians to build an Airbnb business. His ads funnel had a full sales team behind it. His Instagram DMs had one setter, paid on commission, working them about three hours a day. Five to ten new conversations came in daily, message requests got lost, and the DM funnel made nothing.

“I get heaps of DMs, but there’s only one guy on it and it’s very manual,” he told me on our first call. He also worried that messaging new followers would damage his brand. So we started on inbound only, and every DM got an answer in under two minutes, day or night. The AI booked three to four calls a day. Once Jordan was ready, we switched on a personal opener for every new follower, and bookings went to about ten a day.

Jordan’s DM funnel went from $0 to $117,000 in 45 days. The AI replaced his one DM setter, and the setter team on his ads funnel stayed exactly as it was. On whether his team could have built it themselves, Jordan said:

“I think we could have definitely figured it out ourselves, but to the standard at which you guys have made it, definitely not.”

Three more coaches, each with the full interview:

• Kneehab Academy: from four to five DM setters to none in the DMs, and from $45K to $140K a month.
• Sam Johnston: zero setters, two closers, and $27K to $120K+ a month in 90 days.
• Prime Train: from $250K to $350K a month, with the setters moved off the DMs and onto the phones.

All four stories are on the case studies page.