Think about the person on your team everyone asks when the new AI tool does something strange. Now think about who everyone asked a few years ago when a spreadsheet formula broke or a report needed building. I’d bet it’s the same person.
I know this person well, because I have been this person. In Human Resources you see this dynamic often. HR professionals will say they aren’t strong with numbers or technology. The ones who are strong may specialize into functions where the skillset is needed, like People Analytics, Compensation, HR Operations, or HR Technology. The ones who stay in generalist and business partner roles become the go-to for everyone else.
AI is creating a new version of that go-to, and many teams are about to repeat an old mistake. Having one AI power user everyone leans on is a bad long-term plan. You’re putting all your eggs in one basket, and if you aren’t doing anything specific to hold onto that basket, you’re pretty close to losing your eggs.
I’ve seen this movie, and I’ve starred in it
Earlier in my career, a colleague and I were known as “the techy ones” on our HR team. Over time, much of the technical and analytical support for the whole group landed with the two of us.
Since I was the numbers and Excel person, I took on compensation work well beyond the group I supported directly. That extra work became part of what was expected of me, and I was the only one accountable for it.
That is how resentment starts, and it rarely starts with anyone’s bad intent. The skill I had built became a reason to give me more work, and nobody else was asked to build it.
Why it’s the same person, and why AI raises the stakes
The AI power user today is often the same person who was the data go-to a few years ago. This person likely has technical aptitude: an ability to learn new technologies and systems quickly, and enough professional curiosity to invest the time learning.
Owners and managers let one person carry this for a very understandable reason. In the moment, it feels like the easiest solution, and it’s likely the quickest way for the work to get done. I’ll be fair to those managers here, because they aren’t wrong about the speed. What they don’t realize in the moment is that the pattern of always asking this person creates a single point of failure.
With AI, that single point of failure matters more, because the distance between heavy AI use and typical AI use is growing quickly. You can see it across companies. OpenAI’s Enterprise Signals report, published in August and drawn from its own business customers, found that the top 10 percent of companies by AI use were generating 8.3 times as much AI output per active user as typical companies by June 2026, up from 2.6 times in January. That gap more than tripled between January and June. More people have the tools every month, and what keeps pulling apart is how deeply the tools get used.
Section’s July 2026 AI Proficiency Report, a survey of more than 5,000 U.S. knowledge workers, found that 65 percent of managers either set no expectations about AI use or encourage it without holding anyone accountable. When nobody else is accountable, the work flows to the person who already knows how to do it.
What happens when the basket walks out the door
I’ve seen this from both sides over my years in HR. Go-to people leave, often for other opportunities, and when you hear the real reason, feeling unappreciated comes up more often than you’d expect.
A huge gap is left behind. Nobody on the team is prepared to take on what that person carried, and teams can struggle for a long time. The “we miss you” messages that follow a go-to person out the door are kind, and they are also a sign that the knowledge left with them.
If you already have a power user, you have work to do now. Find ways for that person to share what they know, so the skill lives in more than one head. Retention is key, so do not wait for the first signs of disengagement. Retain this person proactively through stay conversations, honest recognition of what they carry, and retention pay where it fits your budget and your pay structure. Keep an eye on their workload as well, because it’s likely higher than you realize.
Hold everyone accountable for building AI skills
Your best bet for the long term is to hold everyone on the team accountable for building AI skills. On a team of fifteen, that starts with a decision.
Decide what adoption looks like first, and how you’re measuring it. Is it daily use? Is it use in a specific use case? Is it completing a training? Determine what success looks like, and then hold everyone on the team to that same standard. Your power user should not be the only person whose AI skills show up in their review.
Encourage people to share their use cases, so the teammates who aren’t sure how to use the technology can see what it looks like in real work. If it’s within your budget, bring in true experts who can help the whole team, not just a few members of it.
This will feel slower at first, because teaching the whole team takes longer than asking the one person who already knows. That time is the price of a team that can keep working when your go-to person takes a vacation, gets promoted, or leaves.
Start with your single points of failure
If you do one thing Monday morning, identify the critical skills on your team and map who has them. A skill only one person holds is a single point of failure. A skill fewer than half the team holds is a risk worth watching. Focus on your single points of failure first.
I built a Critical Skills Matrix to make this easy. It’s free to use, with no purchase needed. Add your people and the skills your business can’t run without, and it shows your single points of failure in red.
Erin Glover, SHRM-SCP, founder of Glover & Co., Newington, CT
Have a single point of failure on your team and not sure where to start? Email me at erin@gloverco.ai.