Digital Employees

What do I have to have ready before you build my AI employee?

By Carol Roderick, PhD  ·  November 17, 2026 Digital Employees

Three things have to come out of your practice before I can build an AI employee: the job written out the way your office does it now, a few examples of that job done well, and one named person who decides how it works. None of it is technical.

An AI employee named Leo writes your LinkedIn posts, and he reports to you

A client you've known for years sends you a message. A friend of hers is looking for an advisor, and she's told him to look you up. You know what he'll find. Your LinkedIn profile has three posts on it and the newest one is from last spring.

You've thought about handing that job to an AI employee. What stops you is not knowing what it will ask of you, or whether you'll end up being the one who writes the posts.

Here is the one I'd build first. An AI employee called Leo writes your LinkedIn posts. He has one job, and one only: the posts that go out under your name, in your words, about the questions people bring you. Leo is the one I run in my own business, so the example comes from work I do every week. Everything he does is in writing. In your practice, he answers to you.

You'll know Leo is working when he meets the goals you set for him. His first goal is reach: how many people see your posts on LinkedIn. Once he's meeting it, you set the next goal up the ladder, and over time that ladder leads to first meetings with new clients. He works to one goal at a time, and you decide when it moves.

Nobody in your practice writes your posts today. That's the work Leo is picking up, and it isn't coming off anyone's desk.

Three things to have ready before an AI employee gets built, in the order I ask for them

I ask for three things, and the order is the part people get wrong. It matters because each one rests on the one before it.

The job written out the way your office does it now. The way the work really happens.

A few examples of the job done well. For a LinkedIn writer, that's posts you'd write again, or the way you explain a question to a client in a meeting.

One named person who decides. The person who says whether the post goes out as it is, or waits.

An example only tells me something once I've seen the steps that produced it. And the person who decides can't settle much until there's a written job and a few good examples on the table.

Almost everything that comes up when you search AI employee onboarding, or an AI readiness checklist like Stratus10's, is written for a whole company getting ready for AI. You're one practice handing over one job. Once those three things are ready, the work follows the steps in AI consulting for small businesses: what to expect.

If the job has never been written down, an AI employee build starts from your sent items

Almost nobody has the job written down, and it doesn't hold anything up. We write it down together, with whoever does that job today, while they do it.

For a LinkedIn writer, the raw material is already there. The answers you've typed to clients over the years. The way you steady a client who rings because the markets dropped. The questions you're tired of answering. The one topic you won't post about, because half your clients would ring you about it.

The rules nobody writes down are the ones that matter most, like which prospects you don't take on, or the question you always ask before you talk about fees. If it isn't said out loud, Leo can't follow it, and that's a correction you'd rather not have to make twice.

Who does what when an AI employee gets built, and how little of it lands on you

Three people are involved, and one of them is me.

You pick the job and you name the person who decides. Your office manager can be that person, and they're often the best fit, because they run the day the AI employee works in. If you'd rather hold the decisions yourself, that works too. Either way it's one name, not "the team".

The person who does the job today sits with me and writes it out, while they work. That's one conversation, not a project.

Then it's my work. I build Leo, I watch what he writes while he learns, and I bring you anything that needs your call. Your systems don't need to be in order first. An AI employee works in the systems you already use, the same way the person doing the job does now.

What you get after an AI employee is built, and what still lands on you

What you get is the job done in writing, every week, in your voice, with the record sitting where you can read it. The posts go out under your name. Opening LinkedIn stops being something you owe yourself.

What still lands on you is reading the work. That doesn't end, and it's also what makes Leo better. You read the posts. Something's off, so you change it and send it back, and what you changed becomes a rule he follows next time.

Once the corrections drop off, you loosen the reins. At the start Leo drafts and you approve every post. Later you can let him schedule the plain ones himself, like the post that shares an article you'd have shared anyway, or a note that you're away that week. The harder ones stay with you longer. You decide how much he does on his own, and you can change your mind.

After the build it moves to ongoing support, and you still have me.

Book Let's Chat AI and we'll go through what an AI employee would need from you

If you're weighing up an AI employee and you want to know what you already have, and what's missing, book Let's Chat AI. It's half an hour. Bring the job as you do it now, even if it only lives in your head, and a post you'd write again, and you and I will go through it.

AI employee questions, answered

What do I have to have ready before an AI employee is built?

Three things, in this order. The job written out the way your office does it today. A few examples of that job done well. And the name of one person who decides what the AI employee does on its own. Say the AI employee is Leo and he writes your LinkedIn posts. That means the posting job step by step, a handful of posts you'd write again, and the name of the person Leo reports to.

What happens if the job has never been written down?

Then we write it down together. That's common. I sit with whoever does the job now and write out what they do while they do it. That includes the rules they follow without thinking about them, like which topics you never post about. Nothing has to be written down in advance.

Who has to be involved in building an AI employee?

Three people. The person who does the job today. The person who decides what the AI employee does on its own. And you, because you pick the job and name that decision maker. Sometimes the first two are the same person. I build the AI employee and watch its work while it learns. The rest of the office doesn't need to be pulled in.

Can my office manager run the build instead of me?

Yes. Your office manager can be the one person who owns the decisions, and they're often the best fit because they run the day the AI employee works in. You still have a part. You choose the job, you tell me and your office manager that the decisions are theirs, and you agree with them on what goes out without anyone reading it first.

Do my systems have to be in order before an AI employee starts?

No. An AI employee works in the systems you already use, the same way the person doing the job does now. There are only three things it can't do: physical work, work inside a tool it can't connect to, and anything by phone. If your client list is in your CRM and your calendar is in Outlook, that's fine. The only question is whether each system has a connection the AI employee can work through, and most of them do.

Carol Roderick, PhD, is an AI adoption consultant who builds digital employees for expertise-led businesses. Each AI employee holds one job and reports to someone on staff. More at https://carolroderick.ca.