What to give to AI
Collecting numbers: sales, stock, cash on accounts and receivables every morning in one report. The first line of customer work: a reply to each lead within a minute, bookings, standard questions. Team summaries: how many leads were handled, how many deals are open, where the delays are.
There is also the owner’s personal routine: the inbox, meeting prep and tracking of assignments. An AI executive assistant takes it on.
What to keep
Choosing directions and priorities. Large deals and negotiations. Hiring key people. Decisions where a mistake is costly. AI prepares the data for these decisions but does not make them.
If the goal is wider than one agent and a report, there is a four-step plan: how an owner can step out of day-to-day operations.
What an owner’s week with AI looks like
The morning starts with a report, not with calls to every manager. Questions to the team come from the numbers: why receivables grew, why lead conversion fell. Once a week there is a meeting on the metrics of agents and people. Once a quarter there is a plan: what to automate next and what it costs.
Typical owner mistakes

Buying a neural network subscription for all staff and waiting for an effect. Starting with the most interesting process instead of the one where the money is. Introducing AI without a metric and, a month later, not knowing whether things improved. Giving the agent decisions instead of routine. Mike’s method avoids these mistakes through the order of steps: numbers, one process, a metric, a decision.
An example week
Monday: the report shows more leads but fewer deals. The question to the head of sales is about a number, not a feeling. Wednesday: the support agent’s report shows a rise in delivery questions, a signal for logistics. Friday: a 30-minute meeting on the metrics of agents and people. The owner takes part in three conversations instead of ten.
Where to start
With one process that has the most routine or losses. A 14-day pilot with the price and the metric fixed in the contract. If you need a person who runs this all the time, there is the AI directorChief AI officer. A chief AI officer is the executive who is responsible for artificial intelligence across the whole company: chooses what to implement and controls contractors, the budget and the risks. format: Mike, 5 hours a week, paid monthly.
What the owner needs to provide at the start
The review takes 30 minutes: you name the processes with the most routine and losses. The pilot needs four things. Access to the system where the process runs, opened in the first three days. A month of sample chats or calls, which the agent learns from. The rules your people follow when they answer: the price list, timelines and what must not be promised. And one responsible employee who spends half an hour a day on disputed cases. The owner does not need to run the pilot personally.


