←  All writing

Everyone on my team uses AI now. How do I know what it does with what our company knows?

Most companies have not lost track of their AI. They never had track of it. Ten questions on what that looks like in a normal week, why banning it fails, and the smallest first step a founder can take this week.

 ·  Field notes  ·  7 min read

FreeThe Starter Kit: fourteen steps, each with the words to paste into Claude.Get the kit

This is the question I hear most from founders in Egypt and the Gulf this autumn. I was asked to answer it in an interview, ten questions, one real example each. These are my answers, lightly cleaned up.

The short version first: 35 seconds, in Egyptian Arabic, with captions.

They never had track of it

The question assumes a company once knew what its AI was doing and then lost it. I don’t think that’s true. In the companies we have sat with, nobody knew in the first place.

All of them are early adopters. Only one of them is building something like a shared brain for its AI. The rest use AI the way most companies do: split by function, each person their own way.

A normal week

The CEO uses ChatGPT on his phone, maybe on his laptop, on a personal subscription. Finance barely touches it, maybe to write a spreadsheet formula. Marketing uses it for content ideas, or for videos if they make videos.

Many of these companies also pay for Copilot or Gemini licences with no use case behind them. Some built automations in Power Automate or Copilot that nobody runs any more. I call that dead AI. The licence is paid, the flow is built, and nothing lives on top of it.

What goes wrong, and who notices

Three things go wrong. The company sees no return on what it spends. It worries about security, about what people paste into tools it does not control. And it feels left behind while everything around it moves.

Everyone notices, each from their own angle. IT sees the risk: personal tools, company data in places it should not be, no rules. The CEO sees the cost and the team’s speed, and gets the security headache from IT on top. The team feels behind, and capped by IT because of the risk.

Nobody sees the whole picture. That is the actual problem.

Why banning it fails

You can’t ban a force of nature. People will find a way around the system. They will put company data in a personal account and break the rule, because the work still has to get done. So give people the right tools, and they stop needing to break the rules.

Picking one tool for everyone fails for a different reason. One tool with no use case is expensive jewellery. The company pays for it, nobody gets value from it, and management ends up answering why it bought it.

One more thing I believe. A company that never finished going digital does not have to finish that first. It can go straight to AI and fill the gaps from there.

What a founder should see on a Tuesday

The work itself. What everyone is working on, read from their actual files. Answers pulled from the company’s context without chasing five people for them. The North Star, the plan and the team’s tasks side by side, so the gaps show.

And planning gets faster. A company can replan next year in a day instead of a cycle that takes months.

I have seen this. With one company in decline, I worked from their profit and loss statements, pulled out the gaps, and built a quarter of work to move them from decline to growth, with a clear return on each change. The changes had nothing to do with AI. AI made them possible.

Where company knowledge should live

In a shared folder. Every AI tool the team uses reads the same folder, so every tool gives the same answer. The folder’s access rules decide who sees what.

A shared drive is something IT already knows how to protect. When accounts get broken into, it is usually down to how the accounts are set up: no two-step sign-in, shared passwords. Turn on two-factor sign-in, set the access rules properly, and the folder is a safe home for what the company knows. Nothing new has to be trusted.

Who owns it

The same people who own it today. The commercial lead owns commercial decisions and facts. The CEO oversees all of it. If the owners are not comfortable with the tech, an AI specialist sits in the room to help.

The part people don’t expect: AI agents can connect to your email. They become colleagues who email you decisions, facts, updates and conflicts. You answer them the way you answer a person, in the tools you already use.

What the team stops doing

Fewer meetings, for sure. Most meetings exist for alignment, and alignment is easier when the context is already written down.

Instead of talking to the person, I talk to their context. I send them work, they send it back with feedback, and the needle moves faster. The company plan is written down too, so I can argue with the plan itself, then send my manager a question. If we still need a meeting, we have one. But that meeting is about strategy, not about who is doing what today.

The smallest first step

One main brain that understands your company, one function, one use case. Test that before anything else.

If that sounds like a lot, take our free kit, give it to Claude, and ask Claude to walk you through it step by step. It will. And if you get stuck, we are around.

If you think this is an IT problem

It’s not an IT problem, and it’s not even an AI problem. How people and AI work together is a business problem. A business that doesn’t use AI well usually also lacks a clear plan, and its written processes don’t match how the work really happens. An IT person can’t fix that.

Put it another way. Is finding your biggest problem and your biggest opportunity an IT job? That is what your team and AI working together are for. Get a return from AI. Not one more tool.

If any of this reads like your company

The starter install is in person or remote. We stay with you until your company brain works, and you keep a map of the gaps, ranked by cost. It starts with a free discovery call: one hour, with Aly.

Book your discovery call