I recently started working with a growing, well-known Australian brand. The brief was refreshingly simple. Help us work out how to think about AI, and help us build out our back-end processes with it, together.
What they did not want was just as clear. They did not want a big consulting firm. They did not want a sales pitch, and they did not want to be sold a product. They wanted one person they could work with: someone to help them navigate the noise, and to build small process improvements alongside them, one at a time, until they had a portfolio of small but high-impact wins.
That brief made me stop and look hard at how I actually do the work.
I use AI the way I am helping them use it
When I sat down to plan the engagement, I looked at what I do with Claude in a normal week. I plan with it. I research with it. It transcribes meetings and summarises them. It drafts documents, builds spreadsheets and surveys, and analyses what comes back.
My part is the thinking. I bring the judgement and the guidance. I push back on the AI when it is wrong or lazy, I challenge its first answer, and I put everything it produces into the context of what actually matters to this client. But the AI does the heavy lifting.
In other words, I work with AI in exactly the way I am helping them learn to work with it. The method and the advice are the same thing.
And somewhere in that planning it occurred to me. In the time it takes a consulting firm to take the CEO out to dinner and present their six or seven figure proposal, I can have actual work delivered. Not a deck about the work. The work.
One person, not a team
One competent, curious person with AI can now do more than a whole big-firm consulting team.
That is not a knock on the people in those teams. Many of them are very good. It is a statement about the model. The traditional consulting engagement was built for a world where research, analysis, documentation and drafting took a lot of human hours, so you needed a pyramid of people to supply them, and the client paid for the pyramid. A lot of that work is now what AI does well, quickly, and for the price of a subscription.
What is left is the part AI cannot do for you: understanding the business, asking the right question, knowing which of twenty possible improvements is worth doing first, and being someone the client trusts. That part never needed a pyramid. It needs one person paying attention.
Consulting, in the big-firm sense, is cooked.
But what if you get hit by a bus?
This is the honest objection, and it is a good one. One of the real benefits of hiring a firm is redundancy. If your lead consultant leaves, gets sick or moves on, there is a team behind them that knows the engagement. Hire one person and you are exposed. Everything lives in their head.
The curious thing is that AI has an answer for this too.
Everything I do for this client is documented, extensively. Every piece of reasoning and every decision is written down live, as it happens, in plain view. Not a tidy report at the end of a quarter: the working record, kept as I go. The reasoning behind a process change sits next to the change itself, so anyone can see not just what was built but why.
That documentation does not cost extra. Writing up the thinking used to be the tedious bit that got skipped when time was tight. Now the AI does most of the writing, so the record is simply a by-product of how the work gets done.
And it lives with the client, not with me. The work sits in two places they own: their own Google Workspace, where their team already works, and a GitHub repository in their account. Everything is there. If I disappeared tomorrow, nothing in my thinking or my implementation would be lost.
That is built-in redundancy, and it is arguably better than what a firm offers. When a firm’s team rotates, the knowledge walks out with them and the next team starts again from the handover notes. Here, the next person picks up the full trail. That next person could be a new staff member, another consultant, or an AI agent. Any of them can read the record and carry on with no loss of continuity.
Small steps, not a transformation programme
The other thing this client understood, and I think it is the wisest part of their brief, is that you do not need a transformation programme to get real value from AI. You need a steady run of small wins. A report that used to take a day now takes ten minutes. A process that lived in one person’s head is now written down and half automated. A survey gets built, sent and analysed in an afternoon.
None of those needs a steering committee. Each one teaches the team a little more about what AI is good for, and the wins compound.
It also helps to have someone with no stake in which tools you buy. I am not a partner or reseller for any AI company, so when I recommend Claude, or ChatGPT, or something else entirely, it is because it is the best fit for the job this month, and the work is built so you can change your mind next year.
So, if this sounds like you
If your business wants to use AI practically, one small step at a time, and you do not want the big-firm pitch, there is another way to do it. People like me.
Send me what you have: a process that eats your week, a question you have been meaning to ask, or just a sense that AI should be helping more than it is. I will help you take the first step, and I will write down everything along the way so it stays yours.