The Four Levels of AI in Your Business

Key takeaways

1. There are four levels of AI in a business: thought partner, assistant, teammate and system. Most owners are at level one and believe they are at level three.
2. Skipping a level exposes your problems. AI doesn’t fix a missing process. It repeats the gap at the same scale.
3. Get organized first. Organized doesn’t mean pen and paper, answers that live only in your head, or work nobody else understands.
4. Keep your data in one place. Three apps that don’t talk to each other can’t give AI context, and QuickBooks is for billing, not your CRM.
5. Stay in the system. Even at level four, someone has to own it, check it and keep it current.

Ask owners how they use AI and many will tell you they hand work to it. Watch how they actually use it and it’s usually a slightly smarter Google. That gap matters more than anything else about AI in a business, because the level you think you are at decides what you try next, and trying the wrong thing next is expensive.

There is a simple way to see where you really are: a four-level scale, from AI as a thought partner to AI as the system that runs your workflows. Here it is, how Y Not You climbed it, and what goes wrong when someone skips ahead.

The four levels

Level one is where almost everyone starts. You download ChatGPT or Claude and treat it like Google: where should I eat, plan my vacation, help me think this through. It is useful, and a little more personal than a search engine, but it isn’t doing anything inside your business.

Level two is AI as an assistant. It completes individual tasks faster and saves your people time. The most common example is an email. You draft it, paste it in and ask for a proofread, then copy the result back into Gmail and send it yourself. You didn’t want to read your own email three times, so the AI did that part.

Level three is AI as a teammate, where you start handing off the repetitive work itself. Instead of copying a draft out and sending it yourself, you let the AI write the email and send it for you.

Level four is AI as the system. It runs critical workflows across the business and expands what your whole organization can handle.

There are no timeframes between the levels. Nobody has to wait two years to reach level two. It is purely a question of how fast you can move. And the value isn’t linear. The more you automate, and the more you use AI to help you automate, the steeper the curve gets.

Most owners aren’t where they think they are

People are really at level one and they think they’re at level three.

The mirror image is just as common: owners who only need level one jump straight to level three, expecting it to fix everything. Francis compares it to skipping levels in a video game. You reach the level three boss before you are strong enough to beat it. The flip side is the player who does beat the boss and has no idea how. An AI producing results nobody understands is as big a problem as an AI producing bad work.

Part of the confusion is the word “assistant.” Some owners hear it and think: go do this, I’m not going to watch. An assistant still needs context.

A teammate works alongside you to finish the task or the project. A system goes further still: it ties the thought partner, the assistant and the teammate together, with parameters set and every task clearly defined. Each level up takes more structure, not less.

How Y Not You climbed the scale

We have lived all four levels. For our first few years, the firm ran the way most small businesses do.

1. NotepadsIn the first year, everything lived in notebooks. Francis still has his.
2. Sheets and docsThen Google Sheets and Google Docs. We talked about Trello and Monday.com. Nothing was automated. There was no real CRM, and no real sense of things like the lifetime value of a client.
3. Shared files with rulesAround the third year, our files and content lived in Dropbox, named by year, month and date. Even that was a starting point.
4. Notion, end of 2022We moved into Notion. For the first two years, it didn’t really work for us. It was one more place to copy and paste information from.
5. All the data in one placeOnce all of Y Not You’s data was in Notion, it became a contextual layer for AI. Today the business runs through it, with clicks of buttons.

Before Notion, we had the setup almost every owner has: cloud storage for files, QuickBooks for the financials, and a Google Sheet listing customer emails. Three apps that don’t talk to each other. A question as simple as what to invoice a customer that month meant digging back through group texts and emails.

There was a human side to the change, too. Francis describes the habit that holds owners back as a lawnmower that only starts if you kick it first: you get so used to the broken way that you stop seeing it. He struggled with the new inputs at first, because he was so used to how we used to work. Now the questions the partners used to text each other about get answered by the AI inside our workspace. Ask it the right way and it can tell you how many times we have worked with a client, because all the context is there.

Get organized first

If you’re even looking to start with an AI transformation in your business, the first thing you have to be is organized. And organized does not mean pen and paper.

It doesn’t mean either of the other two patterns we see most:

• The answer lives only in your head. When someone on your team needs a decision, they have to call you, and you have to be available. That takes time out of your day, every day.
• Someone else does the job, and you have no idea how. You hired someone, the job gets done, and that is all you know about it.

Level three in practice: invoicing

Invoicing at Y Not You used to take two days, at a minimum: two days of copying and pasting from Notion into QuickBooks, one customer at a time, while the three partners traded texts to make sure every invoice was right. Whatever else you were doing, you stopped to keep up with the group chat.

The first fix was a single place for the information, so there were no more texts. Then came the teammate. An automation now takes each invoice from Notion, creates it in QuickBooks and sends it to the customer. Our monthly invoicing went from at least 16 hours to about four.

QuickBooks is for billing, not your CRM

Notion is our CRM. It is where our data lives. Plenty of businesses use QuickBooks as their CRM instead, and they lose track of everything.

QuickBooks is supposed to be for billing. It’s invoicing.

If you don’t have a client relationship manager, you need one. You have more conversations than ever, and if you can’t go back through your own records and quickly find out what a conversation was about, you are losing time. You are losing money, too. Think of the prospect you had a great call with at the start of the month, never followed up with, and later found out went with someone else.

With your relationships in one place, AI can help here as well. We can ask our CRM for a list of hot leads we haven’t spoken to in a few months. If all you have is old text threads, nobody is going back to them.

Marketing needs a backend

Everything on the operations side has to support what you do on the marketing side. You can market like an animal, but if nothing can handle the leads, they never get a touch point, they move on to someone else, and your brand takes the damage.

Marketing is not just putting out videos. It’s having a backend to support your lead funnel when you do start to have people reaching out.

We have watched clients go viral before their funnel was set up, and the attention came to nothing. We have seen clients forced to stop marketing for months while they hired, fired and changed how they work so they could handle the lead flow. Spending money to generate leads you can’t handle is spending it for nothing. That is why our work with clients covers operations as well as content and social media.

AI exposes your problems at the same scale

Everything above speeds up with AI, because work moves faster and more efficiently. That cuts both ways.

It exposes the same exact problems within your business at that same scale.

Take someone at level one, who has only ever used ChatGPT or Claude like Google, and imagine them jumping to level four with an email agent: an AI that runs through the inbox, sorts what came in and writes drafts. They are now trusting an AI system to run an inbox that never had a system. They are hoping it writes emails, proposals and customer replies in a voice they never defined. The agent exposes everything that was missing: no messaging SOPs, no brand guidelines, no cadence for responding to customers. Instead of saving time, they have created a whole new job.

The same thing happens with the ads promising a team of 40 AI employees, one for marketing, one for everything else. We have had clients try it. An AI marketing agent with no strategy behind it produces generic content, and we have watched brands lose their authenticity, their voice and their messaging that way.

You still have to be in the system

A lot of owners want AI to be done-for-you, without being involved at all. Then they look back and wonder why none of it is working, and the answer is that it doesn’t understand the business. It’s generic, with no human touch.

Even if you get to level four and you’ve built a system, you still need to be involved in the system.

Building the system doesn’t mean you can go hang out and do nothing. We see systems break every day because the people who built them tried to go hands-off. Even the people building these AI products tell their users that what they set up six months ago may already be out of date. It is constant maintenance and constant updating.

It is the same principle as running any business. An owner should understand what every person in the business does, especially before hiring for a role, and that means having SOPs and clear guidelines for each one. We hear “I’ll just hire a kid out of college, because they know social media” all the time. Social media is a skill, not something people pick up just by being young. So is AI.

What to do in the next six months

Is there a level five? Not one worth planning for. Level four, AI as the system, is where a business wants to be today: AI either saving you time or producing work for you, proactively, in every part of the business.

Over the next six months, expect the push to come from productivity tools for owners: email agents, text message agents, and new assistants like Meta’s Muse. Our bolder prediction is that within a year or two, the platforms themselves will sell AI agents that run your social accounts, answering comments in your voice and monitoring your DMs. Some of that exists already.

For right now, our advice comes down to two steps:

1. Understand every app you already pay for. Many of them have AI built in, and you may be paying for it without using it. Start there.
2. Then look for the time AI can save you, one workflow at a time.