I saw a job description for a "Head of AI" recently that read like a wish list for a NASA engineer. It asked for a PhD in machine learning and five years of Python, for a firm that mostly needed its email to stop being a mess.
Most owners go looking for a scientist when what they need is a plumber.
The expensive misunderstanding
Hire a full-time developer to "add AI" to your firm and you'll get someone whose job is building things from scratch: new code, new servers, new platforms. That sounds impressive until you realize most of what your firm needs already exists in the tools you pay for. You don't need a new oven. You need someone who knows how to set the temperature and time the roast.
What the role does on a Tuesday
Picture a professional services firm where a dozen people spend half their day moving information from an intake form into a CRM. It's tedious, it's error-prone, and it makes your best people want to quit.
I don't walk in and start building a new CRM. I look at that one process, find where the tools you already own can talk to each other, and build the piece that's missing. I act as a translator: I take the way you actually make money and turn it into a workflow. Then I stay, because the first version is never the last.
What it isn't
It isn't a software salesperson. If someone's first move is selling you a subscription to their own AI platform, that's a vendor, not a partner.
It isn't tech support. I'm not there to fix the printer. I'm there to find where manual work is eating your margin.
It isn't agents everywhere. Anywhere accuracy is contractual, the work runs on a fixed path with a person approving.
Three things to look for
The assessment. They should be able to map your workflows and point to the exact places time is lost. If they can't describe your process better than you can, they can't fix it. The assessment should be yours to keep whether or not you hire them.
Builds on what you own. They should build on your accounts and your tools, so nothing walks out the door when they do.
The adoption plan. Buying software is the easy part. Getting people to use it is where the real work starts. A good partner spends more time with your staff than with the computer.
Try this today
Open your calendar and look at yesterday. Find the one thing that felt like digital manual labor: repetitive, and not needing your expertise. Ask yourself whether you'd hand it off for good if you could. That gap is the job.
David