---
title: When Everyone Picks Their Own AI Tool
description: Letting everyone pick their own AI tool works fine for individual wins. Here's where it starts slowing the shared work down, and what to do about it.
---

[Mickey Baines](https://www.mickeybaines.com/)

[The Book](https://www.mickeybaines.com/book)[Writing](https://www.mickeybaines.com/writing)[Podcast](https://www.mickeybaines.com/#podcast)[Speaking & Consulting](https://www.mickeybaines.com/speaking-consulting)[About](https://www.mickeybaines.com/about)

[Writing](https://www.mickeybaines.com/writing)

# When Everyone Picks Their Own AI Tool

Letting everyone pick their own AI tool works fine for individual wins. Here's where it starts slowing the shared work down, and what to do about it.

November 20266 min read

#### In this piece

#### Share

[in](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.mickeybaines.com%2Fwriting%2Fwhen-everyone-picks-their-own-ai-tool)

⧉

Over the past year I watched something take shape inside the company I was working with. Skills, projects, and artifacts started showing up across the team — some I built, some colleagues built, some genuinely shared and some sitting quietly in one person's account. Nothing had broken yet, and that's the reason it feels worth writing about, because the shape of this problem was visible well before the problem itself showed up.

[Ellucian's most recent higher education AI survey](https://www.ellucian.com/newsroom/ellucians-3rd-annual-higher-education-ai-survey-signals-shift-individual-ai-use) put numbers to a gap most of us can already feel. Ninety percent of higher ed professionals say they use AI. Only 43% say their institution's strategic plan includes an AI focus. Individual adoption has run well ahead of institutional strategy, and it happened almost everywhere at the same time.

Two examples I've built show why that gap matters.

The first is a morning brief. It pulls together my active to-do list, the top emails I need to respond to that day, and my calendar. Before, that lived in three places — Outlook for email and calendar, a notes app for everything else. Now it arrives in one view every morning. If the tool underneath it disappeared tomorrow, I'd feel it and nobody else would. The cost is mine, and I can see it.

![Anonymized sample of a daily morning brief: a timeline of the day, items needing attention, resolved items, a to-do list and calendar](https://www.mickeybaines.com/hs-fs/hubfs/Blog%20Images/mickey-morning-brief-sample.png?width=1584&height=3822&name=mickey-morning-brief-sample.png)

The second was our first marketing dashboard, and it was a different animal. It connected our website analytics and our marketing automation platforms, and once we tied in the CRM, it showed where web leads, conference leads, and people who downloaded content turned into real opportunities, with open, closed, and lost opportunities for every one of those channels. That told us where to continue, accelerate, or pivot in our content and advertising. It was built for our executive leaders, our business unit leaders, and our University Partnerships team, and it replaced people logging into systems they'd otherwise never touch or waiting on me for an update. With the CRM connected, it became the closest thing we had to a shared view of how the whole funnel was performing.

## Before the two dashboards can disagree

That second one is where the tool question gets real. Picture a business unit leader who sees that dashboard, likes it, and decides to build their own version for their team. They prefer a different AI tool. Fine. That's the arrangement most organizations have landed on, and there are good reasons for it.

The question people ask at that point is whether the two dashboards will agree. That assumes the second dashboard gets built at all.

## What happened in our case

In our case, that colleague didn't have the same access to the underlying systems, so they couldn't replicate it. The work came back to me, and I was building and maintaining the same thing twice, in two different tools, for two different audiences. We were just slower, and the slowness sat on me.

The second version is the one that concerns me more. Somewhere — maybe there, more likely at a company with looser permissions — someone has access, but not quite the *same* access, and doesn't know it. They can build a dashboard that looks right even though it's assembled on slightly different data and produces different numbers. How long before anyone catches that, and how many decisions get made off those numbers in the meantime?

That version never announces itself. The first one at least has the courtesy to feel like extra work.

## The part that surprised me

The instinct with all of this is to treat it as a data governance problem. Define the metrics, agree on which system is the source, reconcile the numbers when they disagree. We did that. The cost I kept running into sat somewhere else.

[Zylo's 2026 index](https://zylo.com/blog/ai-in-workplace) found the average company now runs seven generative AI applications, and 60% of IT leaders say they can't see every one in use. That's the environment most mid-sized organizations are operating in right now, whether or not anyone has said so out loud.

What that produces is one of two outcomes. Either the shared views don't get refreshed evenly and people start making calls off stale information, or the work piles onto the few people who can actually do it, and those people slow down. Either way the organization's AI progress slows.

Every one of those tool choices made someone faster on their own work. Put together, they're what's slowing the shared work down.

[BCG's 2026 workplace survey](https://www.bcg.com/publications/2026/ai-at-work-why-strategy-matters-more-than-tools) found something that fits this uncomfortably well: employees with a clear strategy but limited access to tools outperformed employees with strong tool access and no direction.

## Where I think this goes

I don't read this as an argument that experimentation was a mistake. The early phase of AI adoption is learning and experimentation, and that phase genuinely benefits from people using whatever they're most comfortable with. Having the freedom to pick my own tool has been freeing as an individual user, and I'd say the same for most of the team I worked with.

But that phase ends. Once you've learned enough to start prioritizing — which use cases matter and which ones the organization is going to depend on — the tool becomes one of the things you're prioritizing. By then, replicating and sharing successes is the whole game, and the tool is part of that whether or not anyone has declared a standard.

For mid-sized and smaller teams, some form of convergence stops being optional. The time savings stay capped otherwise, because they never get past the person who built them. Larger organizations have IT governance that will grind this out eventually. Most of higher ed, and most of the companies serving it, don't.

There's a version of convergence that's total, and a version that's scoped to specific use cases — anything touching company data, or feeding a view more than one person depends on. Both work. What matters more is that leaders actually know what's being built while it's being built, especially on teams running different tools.

For anyone who reads this and concludes the answer is to standardize and move on: getting everyone onto one tool doesn't finish the job. We were seeing individually created skills that were genuinely good and lived entirely inside one person's account. Same tool, same company, and the good work still doesn't travel. Those have to be reviewed, and where they're worth it, turned into a standard. Otherwise the successes stay stuck with single users and the organization never actually compounds anything.

By the time I left, the dashboard was finished, but we were still arguing about which skills belonged to individuals and which belonged to the company. If your team is about six months from that same conversation, I'd schedule it now, while the list is short enough to sort through in an afternoon and before the person who built the best one walks out the door with it.

MB

### Mickey Baines

I help higher ed leaders clarify and align their answer to strategic priorities and then design the path that will get them there.

### Get the next piece in your inbox

New writing, podcast episodes and upcoming sessions.

## Keep reading

[All writing →](https://www.mickeybaines.com/writing)

### [November 2026 Intelligence Is the AI Buzzword in 2027. Here's What's Driving It. Intelligence is about to show up in product names, job titles and department names. Here's what's behind the word, and what it means for your data.](https://www.mickeybaines.com/writing/intelligence-ai-buzzword-2027)

Coming December 2026

### “Just implement best practices.”

What this sentence costs, and why AI is making it expensive.

Speaking & Consulting

## Would you like your team to see the future a little more clearly?

I connect with audiences of every size, from a conference keynote to a small leadership group.

[Start a conversation →](https://www.mickeybaines.com/speaking-consulting#start-a-conversation)

Mickey Baines

© 2026 Mickey Baines · [Privacy](https://www.mickeybaines.com/privacy)

```json
{
  "@context" : "https://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Mickey Baines",
    "url" : "https://www.mickeybaines.com/writing/author/mickey-baines"
  },
  "dateModified" : "2026-11-14T18:06:27.176Z",
  "datePublished" : "2026-11-14T18:06:27.000Z",
  "headline" : "When Everyone Picks Their Own AI Tool",
  "mainEntityOfPage" : {
    "@id" : "https://www.mickeybaines.com/writing/when-everyone-picks-their-own-ai-tool",
    "@type" : "WebPage"
  },
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject",
      "url" : "https://www.mickeybaines.com/hubfs/spark-scale-cover-1.png"
    }
  }
}
```