You Don't Have an AI Problem. You Have an Authority Problem.

I run a diagnostic tool that scores companies across five dimensions of AI readiness: how widely you have adopted AI, who gets to make decisions about it, how deeply it fits into your actual work, how fast those decisions move, and whether the results show up in your financials. I have now seen enough assessments to know which dimension predicts the overall score most reliably.

It is not adoption. Most companies are already using AI. McKinsey’s 2025 State of AI survey found that 88 percent of organizations report regular AI use in at least one business function, up from 78 percent the year before. The tools are in the building. That is not the bottleneck.

The bottleneck is who gets to say yes. Specifically: who is allowed to approve an AI initiative, how many people have to agree, and what happens when two of them disagree.

The sign-off chain

Count the number of approvals an AI initiative needs between the moment someone identifies a use case and the moment a production budget is released. I call this decision distance. In the mid-market companies I work with, the number is usually four to seven. In Fortune 100 companies, it is twelve or higher.

The number matters because every sign-off is a potential stall. And AI initiatives stall differently than other technology programs. A CRM migration stalls because of data quality or change management. An AI initiative stalls because the person three levels up does not understand what it does and is not compensated for the risk of approving it.

RAND Corporation research found that 80 percent of enterprise AI projects fail to deliver promised business value. Of those failures, a third are abandoned before reaching production. The conventional reading is that the technology was not ready. The data suggests the organization was not ready, and the specific part that was not ready was who gets to decide.

Governance is not the same as authority

Companies often respond to AI anxiety by building governance. A steering committee. A responsible-AI policy. An intake form. A quarterly review cadence. This is reasonable and necessary. It is also not the problem I am describing.

Governance tells you what the rules are. Authority tells you who can say yes within those rules. The companies I see struggling have plenty of governance. What they lack is a clear, short path from “this workflow should change” to “here is the budget to change it.”

ISS Corporate’s 2026 governance report found that only 22 percent of S&P 500 companies have disclosed board-level oversight of AI. The number for the mid-market is almost certainly lower. That means in most companies, AI decisions are being made by people who do not have explicit authority to make them. The work happens anyway, in gaps between job descriptions, and the results are invisible to the people who control budget.

What the authority problem looks like in practice

Here is the pattern I see repeatedly. A director-level leader identifies a workflow that AI could meaningfully improve. Donor communications at a nonprofit. Renewal quoting at an insurance company. Proposal generation at a professional services firm. The leader does the research, maybe runs a pilot on a free tool, and gets promising results.

Then they try to scale it. They need budget. They need IT to approve the vendor. They need legal to review the data handling. They need their VP to sponsor it in the quarterly planning cycle. The VP needs to explain it to the CEO, who needs to feel confident enough to mention it to the board without sounding naive.

Each of those conversations introduces a new opportunity for the initiative to die quietly. Not because anyone says no. Because nobody says yes fast enough, and the window closes, and the next quarter’s priorities take over.

Grant Thornton’s 2026 AI Impact Survey found that three out of four organizations admit their governance has not kept pace with AI adoption. The mechanism is not that governance is too strict. The mechanism is that authority is too diffuse. When everyone has to agree, nobody has to decide.

The mid-market version of this problem

Large enterprises have the authority problem for structural reasons: complex org charts, matrix reporting, multiple P&L owners. The mid-market has it for a different reason, and the reason is more fixable.

In most mid-market companies I work with, the senior team is small enough that authority could be clear. Five to fifteen people make every important decision. The problem is that AI does not fit neatly into any one person’s domain. It is not purely an IT decision. It is not purely an operations decision. It is not purely a strategy decision. So it floats between all three, and each person assumes someone else is driving.

The fix is specific and unglamorous: name one person who owns AI outcomes. Not AI governance. Not AI policy. Outcomes. Give that person a budget, a kill threshold, and a reporting line to the CEO. The RSM 2025 Middle Market survey found that 91 percent of mid-market executives have adopted generative AI, but only 25 percent have it fully integrated into core operations. The gap between those two numbers is an authority gap.

The test

Here is how to know whether you have an authority problem. Answer three questions honestly.

If someone on your team found an AI tool tomorrow that could save $200,000 a year in a single workflow, how many weeks would it take to get a production budget approved? If the answer is more than six weeks, you have an authority problem.

If that same person’s pilot failed after eight weeks, would anyone formally kill it and reallocate the resources? If the answer is “it would probably just quietly stop,” you have an authority problem.

Is there one person in your company whose job title or written mandate includes the phrase “AI outcomes” or “AI value capture”? If the answer is no, you have an authority problem.

The companies that close the gap between AI adoption and AI value are not the ones with better tools. They are the ones where someone specific is authorized to make the call, funded to execute it, and accountable for the number.

How RLK Can Help

My AI Diagnostic scores how decisions get made in your company alongside four other dimensions and shows you exactly where the friction sits. If the score confirms what you already suspect, the AI Strategy Session is a focused engagement that maps authority to outcomes and builds the decision framework your team is missing. Get in touch.

Sources

• McKinsey, “The State of AI in 2025: Agents, Innovation, and Transformation”
• ISS Corporate, “Artificial Intelligence and Governance: Is 2026 a Tipping Point?”
• Grant Thornton, “2026 AI Impact Survey Report”
• RAND Corporation, Enterprise AI Project Outcomes Research
• RSM, “The Real Economy: Middle Market AI Adoption Survey 2025”
• Harvard Law School Forum on Corporate Governance, “US AI Oversight Through Three Lenses”