Automate Is Step Five.
Elon Musk runs a five-step business algorithm. Here is the order, and why it changes how AI lands in your business.
For 16 years I ran a gym, and there were mornings I needed 7 tabs open just to answer one question about my own business.
The question would be something simple. How many active members do we have right now?
Seven tabs later, here is what I had. The booking platform with one count. The payment processor with another. A spreadsheet our CSM updated by hand. A QuickBooks export from the bookkeeper. An email list parked in a fourth tool, last cleaned a year before. A Google Doc the head coach kept open, because nobody trusted any of the official platforms to be current. And of course the one primary tool, that could never give us the correct answer, even though that is what it promised.
I did not have an answer at the end of those 7 tabs. I had six partial answers that disagreed with each other.
That was the business I was trying to grow on top of.
It is not just gyms
It is the same pattern almost everywhere right now. Tool sprawl. Login fatigue. Data strung across the organization like birthday streamers hung at random angles, half of them falling down. By the time you have logged into the six systems you need to answer a basic question about your own company, the question has changed.
The reason most of us want AI right now is the same reason we bought all those tools to begin with. Each one promised to fix a problem. Each one did, a little. Each one added a login. Each one added a data island.
You can’t put AI on top of a mess
Well, you can. It will just give you a faster mess.
If your customer list lives in three places that disagree, an AI agent will produce three different answers depending on which place it reads. If your team has 40 logins and no single source of truth for what closed yesterday, your agent will sound confident while it hallucinates, because the underlying ground truth is already a hallucination.
AI amplifies what you already are. I have said that for two years. It is true for people. It is also true for businesses. A simple, organized, decided business gets multiplied by AI. A bloated, scattered, undecided business gets noisier.
The owners who are winning with AI right now did not start with an AI strategy. They started with a deletion strategy. They cut tools. They consolidated platforms. They picked a single source of truth for each domain and enforced it. Then they brought AI in, and AI worked, because there was finally a clean substrate for it to work on.
There is a name for that approach, and it does not come from a small business consultant.
Elon Musk has a five-step process. Automate is step five.
When looking at the world of business, and specifically how to run & automate major businesses, Elon is the king. He runs the same algorithm at SpaceX and at Tesla, and he talks about it openly. Five steps, in this order:
Question every requirement. Who said this part of the business had to exist? Why? Is the reason still true?
Delete the part or the process. If you do not end up adding back at least ten percent of what you cut, you did not cut hard enough.
Simplify and optimize what is left. Only after deleting. You do not optimize something that should not exist.
Accelerate cycle time. Move faster on what remains.
Automate. Last.
Read that list one more time. Notice where automation sits.
Most of us want to start at step five. Buy the AI tool. Bolt it onto the way the business already runs. Hope it untangles everything from the inside out. Musk does the opposite. He spends almost all of his time on steps one through three. Only when the operation is at its simplest possible version does he automate any part of it.
I am not telling you to run your business like Elon Musk. I am telling you the order matters. Question, delete, simplify, then automate. If you skip the first three steps, the AI you bring in is just step five with nothing underneath it. A faster mess.
That is the algorithm I run in week one of every engagement. It is also why an off-the-shelf AI rollout almost never delivers what the demo promised. The demo skipped steps one through four.
The work nobody wants to do
Inventory every login. Every platform. Every spreadsheet. Most stacks I have audited carry between 15 and 40 active tools. The gym did. Half of those tools turn out to be doing a job another tool already does. A third are doing nothing at all. They are paid subscriptions for software somebody on the team forgot was running.
Pick one system of record per domain. One CRM. One project tool. One place where customer notes live. One folder for documents. One.
Then deprecate the rest. Put a hard date on the calendar. “Nobody opens [old tool] after May 15.” Move what matters. Let the rest die.
Centralize your data. If your agent reads the customer list, it should read one customer list. If it reads the pipeline, it should read one pipeline. The agent does not need the prettiest interface or the most expensive tool. It needs a single, clean, current source.
This is the unglamorous work. It is also the work that decides whether your AI investment pays back in 90 days, 1 year or never.
Every login is a tax
Watch your team for a day. Count the times someone asks “where did I save that?” or “what was the password for that one?” or “who has access to that platform again?” Then count the seconds lost on each one. Then count the small reset of attention each switch costs. Multiply by your team. Multiply by a year.
That number is brutal. And it is the tax you pay before AI ever shows up.
Now layer AI agents on top. Each agent is its own tool, with its own interface, its own login, its own dashboard. If you do not simplify first, AI just becomes the 41st thing on the list.
The phrase I have been using with clients lately is switching fatigue. It is not just slow. It compounds. Every platform your team has to remember the location of is one less platform’s worth of energy they have to actually do the work.
Simplicity is not the constraint on AI. Simplicity is what lets AI work.
How to start, this week
Open a blank document. List every tool, platform, login, and subscription your business runs on. Three columns. Tool name. What it does. Who actually uses it.
Most of us stop around 25 entries and realize we cannot remember what half of them are anymore.
Then three questions, in order.
What job does this tool do?
Which of these jobs is duplicated by another tool on the list?
Which of these tools, if I cancelled tomorrow, would nobody on the team notice for a week?
The last column is your starting kill list. Cut it. Do not migrate. Just cut.
Then do it again next week with the next layer down.
In a month, your stack is small enough that AI has somewhere clean to live. In two months, your team is not drowning in logins. In three, the agents you bring in pay back their cost in week one, because the data finally agrees with itself.
The main reason I prefer to do a hardware AI installation along with the AI Agents, is it gives each owner a chance to start fresh. A fresh computer, zero files, and then we give the AI only what it needs to be successful. What you find out, is that it only really needs a very small handful of documents and tools. And those documents and tools are actually the only things your company actually needs to hang on to.
Go ahead and delete all the project photos and proposals from 10 years ago. Sort your Google Drive by “date last modified” and cut everything that hasn’t been touched for over 3 years (double checking it’s not taxes or something pressing like that).
My analogy for this is it is like downsizing from a 5,000sqft 5 bed, 3 bath house to an empty nester 2,000sqft 2 bed, 1 bath ranch home. Purge purge purge, and you will feel so light, mobile and agile afterwards that you’ll wish you’d done it 5 years ago.
The clean version is in there
I am not anti-tool. I am anti-sprawl. There is a clean version of your business hiding underneath the platform pile, and most owners cannot see it because they are too tired from the daily logins to step back and look at the whole thing.
Question, delete, simplify, accelerate, then automate. The order is the whole point. Most small business AI projects start at the end of the list and wonder why nothing changes.
If you want help running the inventory or picking your starting kill list, reply to this email. I run this exercise with every Tier 2 and Tier 3 client in their first week. It is not the glamorous part of the engagement. It is the part that makes everything after it possible.
Dublin, Ohio. Coffee in hand. Four tabs open this morning, on purpose.
Jeff