The fix is structure, not a script you paste into every message. I build the follow up around real details about that client, then let AI handle the sequencing and timing, never the sentence itself. The moment AI writes the actual words a client reads, the message reads generic, because it is.

The robotic feeling clients describe is not a tone problem. It is an information problem. A follow up that says “just checking in” could have gone to anyone on your list, and clients know it the second they read it.

What Makes an AI Generated Follow Up Feel Robotic in the First Place?

It feels robotic when the message could apply to any client, not this one. AI defaults to smooth, generic phrasing unless you feed it specific detail, so a follow up written from a blank prompt reads like a template every time.

The mechanism is simple. AI compresses whatever you give it. Give it nothing specific and it compresses down to nothing specific. Give it the real detail and the output holds the detail.

What Should the AI Actually Be Doing in My Follow Up Process?

AI should own timing, sequencing, and the first draft pull from your notes, never the final send. It reads your CRM notes, drafts a message using specifics you logged, and queues it. You review and send.

In my own business, the sequence looks like this. I log three lines after every call, what they said, what they are deciding, what I promised. AI turns those three lines into a draft at the right interval, three days, then eight, then whatever cadence fits that relationship. I read every draft before it goes.

This is the same split I walk clients through when they ask about which processes to systemize first. The repeatable mechanics get delegated to the system. The judgment about what actually matters to this client stays with a person, at least until the draft has proven itself for months.

Where Do Founders Usually Get This Wrong?

Most founders automate the whole message instead of automating the trigger. They let AI write and send without a human read, which is how a client gets a follow up that ignores something they said on the call last week.

An owner sets up a tool, feeds it a generic prompt, and turns the whole sequence loose unattended. It works for a month, then a client mentions the tone shift, or worse, gets a message that contradicts what was actually discussed.

The Phase Check exists partly because of this pattern. Owners think they have an AI problem when they actually have a documentation problem. If your notes after a call are thin, no AI tool fixes that. It just moves the thinness downstream to the client’s inbox faster.

How Do I Know If My Follow Ups Are Actually Working?

Track reply rate and time to reply, not send volume. A follow up system that sends on schedule but gets ignored is not working, it is just running. Reply rate tells you whether the message still sounds like a person who remembers the client.

I run this exact review in my own business every Friday, alongside the rest of my weekly scorecard. If reply rate drops two weeks in a row, the first thing I check is whether the notes feeding the AI draft got thinner, not whether the tool needs replacing.

Most operators blame the tool when the real issue sits one step earlier, in what gets logged after every call. Fix the input and the output stops feeling automated, because in the parts that matter, it is not.

Frequently Asked Questions

Can AI write client follow ups that sound like me?

Yes, if you feed it your own past messages and specific call notes rather than a generic prompt. The tone comes from the source material you give it, not from asking it to “sound friendly.”

How often should I send AI drafted follow ups?

Match the cadence to the relationship stage, typically every three to five days early in a decision and every one to two weeks once it stalls. A fixed schedule for every client is one of the fastest ways to sound automated.

Should I let AI send messages without me reviewing them first?

No. Review every draft before it sends, at least until you have months of proof the drafts hold up. This is the same build first, trust second sequence I use for any operator process before removing myself from it.

Take the Next Step

If you are not sure whether your follow up process is a tool problem or a documentation problem, take the free Phase Check. It takes a few minutes and I read every result myself. And if you would rather talk it through directly, here is how my coaching works.

Anthony Spitaleri

Business Performance Coach

anthonyspitaleri.com

About Anthony Spitaleri

I’m a Business Performance Coach. I work with owners on the decisions that let a business run without them: the first key hire, a sales team that doesn’t depend on the owner, and the numbers worth watching each week. I worked through those same decisions at a professional services firm I helped grow from a small team to more than a hundred people across two countries, where I found and fixed the gaps in sales so it could scale, built the hiring process and handled the finances. Start with the free Phase Check, or read about working with me.

Sources referenced: SBA Business Guide, IRS Small Business and Self-Employed Tax Center, US Census Bureau Business Trends and Outlook Survey, which tracks AI adoption across US firms by size and sector.