Stanford studied enterprise AI deployments and found that only six percent of companies had data that was actually ready for AI. If you are a founder, that number should change who you hand your AI plan to.
Most founders I talk to explain a stalled AI effort in one of two ways. The models are not good enough yet, so we wait for the next release. Or we do not have the right people, so we go and hire them. Both stories are comfortable, because both put the problem outside the building.
The Stanford figure puts it back inside. The vast majority of companies asked AI to help and had nothing coherent to show it. That is not a model gap and it is not a talent gap. It is a data gap. And here is the part I want every founder to hear: closing it is not an IT project. It is a leadership decision.
Count Where Your Business Actually Lives
Start with one question. Where does your business actually live? Not where it is supposed to live. Where it lives today, as the work gets done.
When I sit down with companies, the honest answer is usually something like forty disconnected places. That is my estimate, not research, so count your own and see how far off I am. The list tends to look like this:
- Customers in the CRM, except the newest ones, who live in a salesperson's inbox.
- Pricing in a spreadsheet, and the real pricing, the discounts people actually give, in a different spreadsheet one person owns.
- People data in an HR tool. The books in accounting software.
- Contracts in a shared drive, project status in a task tool, complaints in a support inbox, recordings in a video tool.
- The decisions that explain why the company does what it does, buried in chat threads and meeting notes nobody will read again.
Every one of those tools was a reasonable purchase. Together they are not a company. They are a scatter.
AI Can Only Help With What It Can See
Now ask AI to help. Help us sell more. Help us spot the clients who are drifting. The only honest answer back is a question: with what?
A model does not know your clients. It does not know which ones have open deals, which ones you have not spoken to in a month, which ones are late on an invoice. If clients live in one tool, deals in a second, meetings in a third and invoices in a fourth, the AI sees four strangers and no company.
So it does what it can. It summarizes an email. It drafts a paragraph. Useful, but a toy. And the founder concludes, wrongly, that AI is overhyped. I made the same point about undocumented know-how in Your AI Isn't Failing. Your Tribal Knowledge Is Locked Up.: the AI was never given the one thing it needed.
The six percent is not a budget club or a talent club. It is a data club. Its members are the companies that made their facts visible in one place.
Plenty of them are small. A small company has fewer facts and fewer tools, so it can get to one home faster than a conglomerate ever will. And you can hire a brilliant machine learning engineer, but if your data sits in forty places, you have paid a premium for a very expensive person to do plumbing.
Why IT Can't Get You Into the Six Percent
The entry fee is one decision: one home for every fact. Every customer, deal, meeting, invoice, person and price in one database, as one connected set of records. Not forty tools wired together through connections that break on a Tuesday.
The decision comes with a rule the whole company lives by: if it is not in the system, it did not happen.
This is where most founders hand the job to the wrong person. IT can pick a database. IT can migrate records. What IT cannot do is tell the top salesperson he gets no credit for the deal sitting in his inbox. It cannot tell a senior manager that the discount she agreed in the hallway does not exist until it is on the record. It cannot decide that a call made in a chat thread is not a decision until somebody writes it where the company can see it.
Those are leadership calls. The software is the easy part. The rule is the hard part. Companies that buy the database and skip the rule end up with a forty-first tool. Companies that adopt the rule end up with a company that can see itself.
- One decision. The business picks a single home for its facts, and leadership says so out loud.
- One rule. If it is not in the system, it did not happen. Said to the whole company, and meant.
- One database. Every record connected, kept current because people treat it as reality.
What One Home Looks Like in Practice
At Edge8, my company, we run the business on a system we call the Company OS. Open it and you see every company we work with on one screen. Click into one and the people, the open deals, the meetings and the invoices are already there, as one record. Not five tools with a dashboard on top. One record, because it is one database.
Because everything is already connected, a plain English question can be answered across all of it. Here is one I asked: which clients have open deals but no meeting in the last thirty days?
In the example I use, that list used to take three people and two days. One pulls deals from the CRM, one exports the calendar and matches meetings by hand, a third reconciles the two and argues about which contacts count. By the time it arrived, it was stale. In one home, the AI answers it instantly, because it can see deals and meetings together.
Then the two questions I ask of every AI claim. What did the AI do? It answered a question across the whole database and handed me a list. What does the person still decide? Which clients to call, in what order, and what to say. The AI does the seeing. A leader does the deciding.
The First Thing a Certified AI Officer Owns
When founders ask me what an AI Officer should do first, they expect an answer about tools or agents. My answer is this decision. Before any agent, any automation, any pilot, someone has to own the one home and the rule that keeps it true.
That person needs authority, not just access. They redesign workflows so facts land in the system as part of doing the work, not as homework afterwards. They hold the line when a senior person wants an exception. They work with engineers on the plumbing, but the ownership sits with them. It is the same pattern behind why so many AI pilots stall: the technology is ready long before the leadership is.
What to Expect Your Leaders to Be Able to Do
If you are deciding which of your leaders should carry AI, here is the bar I would hold them to. A leader who is ready can:
- Produce the honest count. Name every place the business actually lives, without cleaning up the list. The length of the raw list is the point.
- Make the call on one home. Pick it, explain it, and stop the forty-first tool from being bought.
- State the rule and enforce it. Including with the people who are hardest to say it to.
- Redesign the workflow, not just the storage. So data entry is part of the work, not a chore after it.
- Ask the cross-cutting question. Put a plain English question to the whole database and get a usable answer.
- Draw the line between AI and judgment. Say exactly what the AI did and what a person still decides.
None of that is coding. All of it is leadership. It sits right next to the three skills to lead AI I look for in anyone who will run AI inside a company.
What to Do This Week
Three steps, in order. First, count the places your business lives and write every one down. Second, make the decision: pick the one home, then say the rule to the whole company. Third, give it an owner, someone with the authority to keep it true.
That third step is where most founders get stuck, because they do not yet know which of their leaders can carry it. That is what AI Officer certification is built to show you. The challenges run on your own data and workflows, not toy examples, and within 90 days you know who is best equipped to lead. If you want to find out which of your leaders can make this decision and hold this rule, start there.