AI Consulting | Start From the Right Place

Everyone tells you to “use AI”; nobody shows you where to start. Here is how I define AI consulting: we go through the repetitive work in your business one item at a time, separate what today’s tools can genuinely take over from what they cannot, pick one of them, build it, get your team using it, and then measure whether it actually works. Not a tour of tools — a system built on top of your own operation.

This page covers the service itself. If you want to know what a consultant actually does hour to hour, see What an AI Consultant Really Does; for how the fee is set, AI Consulting Fees; and for whether to work with one person or an agency, AI Consultant or Agency.

What the Engagement Covers

Four parts. You do not have to take all of them.

• Assessment. We list everything repetitive in your business: whose hours go where, and where the output of that work ends up. You end up with two lists — what can be handed over today, and what cannot. The second list matters as much as the first, because it shows you the money you would have wasted.
• Pilot build. We pick one single task from the list and build it. It might be a system that answers from your own documents, a sorting layer for incoming email, or invoices turned into a spreadsheet. The reason for picking only one is below.
• Getting the team to use it. Something built and unused is the same as something never built. We write down who asks what and how, and I walk the team through it in a one-hour session.
• Maintenance and monitoring. Models change, prices change, your process changes. Monthly checks on what was built are a separate piece of work — details on AI System Maintenance and Monitoring.

How the Process Runs

1. Intro call. I listen to what you do and what wears you down. I will not try to sell you anything on this call; if the work is not a fit for me, I say so here.
2. Assessment session. We walk through your processes one by one. The output is a written list: which task, estimated gain, estimated build time.
3. Scope and price in writing. What will be done, how long it takes, what it costs — in writing, in your hands.
4. Build. Most jobs finish between 2 hours and 2 weeks — what moves that range is covered in How Long Does an AI Project Take.
5. Handover. What gets built runs in your accounts, on your data. It keeps working after I step away.

The Mistake I See Most Often

Picking the most complicated process for the first automation. The “if we’re doing this, let’s start with the hardest part” reflex comes up in nearly every intro call, and it is the single most common reason these projects fail.

Here is why: a complex process carries many exceptions, not all of them known upfront. The build drags on, the team sees results late, and the verdict settles in — “this doesn’t work for us.” Meanwhile, that same business has a plain copy-paste task eating four hours a week. It takes two days to build, the team sees the benefit immediately, and they bring you the second task themselves.

I recommend the opposite: start with something boring, repetitive, and low on exceptions. Do the hard one after the team trusts the tool.

Who This Is Not For

Saying this upfront saves us both time:

• If your process is not written down anywhere and you have no intention of writing it. AI cannot guess at work that was never described. We can write the process together during the assessment, but come knowing that is real effort too.
• If your data is sitting on paper. Scanned documents work; a folder in a drawer does not. I do not do physical archive scanning — that is separate work and belongs with someone else.
• If you expect “build it once, never look again.” An unmonitored system quietly breaks within a few months. I have seen it happen; it will happen to you too.
• If you are expecting video production or design work. That is not what I do.

An Observation From My Own Operation

I followed exactly this order in my own business. Today I manage dozens of Google Ads accounts and GA4 properties, plus a set of Meta ad accounts. At that scale, running routine checks by hand is not possible.

But I did not start by automating the hardest part — campaign optimization decisions. I started with the most boring part: is there budget drift in any account, has conversion tracking quietly broken, which keyword is burning money. These are low-exception, clear-rule tasks. I wired those scans to Claude-based agents; I handle the exceptions that need judgment myself.

The concrete difference showed up in one place: conversion tracking breaks without throwing an error, and goes unnoticed for days. Under manual checks, you find out when the client says “we’re getting sales but the report doesn’t show them.” With automated scanning it gets caught the same day. The gain is in the ad budget that does not get wasted.

You Can Hire Me for This

You can hire me for this: I work remotely, billed hourly. The assessment session can be bought on its own; you decide about the build afterwards. Why I do not quote a fixed package price is explained on the fees page — a price given without knowing the scope misleads either you or me.

If you are not sure where to start, an AI Readiness Audit may be the better entry point. Write to me from the contact page.

Frequently Asked Questions

What exactly does AI consulting cover?

Four parts: assessing repetitive work, building one selected task, getting the team to use it, and maintaining what was built. You do not need all four; most engagements start with an assessment session and continue with a single pilot build.

We have no technical staff. Does this still work?

It does. Most of what I build sits on tools that require no technical knowledge — the person using it types a question and gets an answer. Technical staff only become necessary when you want a private setup running on your own server, and that is a rare requirement.

Will my data leave the company?

It depends on the tool, and we decide that together. With off-the-shelf tools, your data falls under that vendor’s terms; if sensitive documents are involved, we read the Team/Enterprise terms together. If the data has to stay entirely with you, the build is planned accordingly. More detail: Connecting Your Company Data to AI: A Roadmap.

When will I see the first result?

If the pilot task is chosen well, usually within the first week. That is precisely why I recommend starting with something simple and repetitive — the team seeing the benefit early is what decides whether there is a second step.