This website uses cookies

Read our Privacy policy and Terms of use for more information.

Last Thursday I spent the day at a private equity event, doing back-to-back meetings with PE firms and the folks running their portfolio companies. In one of them I pulled out a sketch I'd drawn in my notebook of how we see AI playing out for mid-market companies.

I've walked about a dozen leadership teams and PE firms through some version of that sketch over the last 2 weeks, and it's landed every time. The part that gets the biggest reaction: getting faster with AI isn't going to protect your company.

1. Productivity is table stakes now

Most of the AI work I see right now goes toward getting faster. Meeting notes that write themselves, first drafts, summaries, reports nobody has to build by hand anymore. That's good work and you should do it. BUT everyone else is doing it too, so a year from now it's just the cost of staying in business.

The way I see it, you're now competing with Claude and OpenAI and everyone building on top of them, along with the other companies in your industry. Your industry is going to get disrupted at some point.. and ideally you're the one doing the disrupting.

2. Last year I told you to save the big bet for later

About 14 months ago I wrote about 3 horizons for AI (efficiency, differentiation and disrupting your industry). My advice then was to map your ideas by effort/impact, start with the highest impact, lowest effort items and document the rest for later.

I've changed my mind on the "later" part. The tools got good so fast that later shows up a lot sooner than it used to, and the companies waiting for the efficiency work to wrap up before they think about disruption are going to find someone already did it to them.

We also renamed the horizons in plain language so they're easier to talk about in a leadership meeting. We call them the 3 frontiers now:

Three retro print cards: a pager, a boombox and a joystick, one for each frontier.
  1. Streamline. Use AI to create efficiencies and speed up the business you run today.

  2. Stand out. Use AI to stand out against your competitors.

  3. Disrupt. Be the disruptor in your industry.

3. Put a few of your best people on frontier 3 now

You don't have to finish one frontier before you start the next. Work on all 3 at once. Most of the early effort will land in Streamline (frontier 1), and that's fine, because it's where the foundation gets built and where the first results show up. It pays for what comes next.

Give frontier 3 (Disrupt) real time and attention as soon as you can, though. A few of your best people, a few hours a week and a deadline. I read a great line last week.. "Small teams with a deadline ship." The same post points out that Google's NotebookLM started as a prototype from 4 or 5 people working part-time for 6 weeks.

A good first question for that team: if a startup with AI and none of your legacy systems came after your best customers, what would they build? Then build a small version of it yourself before they do.

4. You still have to go through the stages

None of this means skipping the hard part. Last week I wrote about the 4 stages every company goes through with AI, and we've tightened up the names since then:

Cassette label reading 4 STAGES, with four blocks of bars stepping up from left to right.
  1. Set the direction. Leadership makes AI a priority and decides where to invest first.

  2. Lay the foundation. Get the rules out of people's heads, map the workflows, set guardrails, clean up the data that matters and put AI in everyone's hands with real training behind it.

  3. Build your AI operating system. One view of your business (for non-technical folks, think one place where your tools/docs/data come together), connected to your systems, with AI underneath that understands how your company actually runs.

  4. Add AI teammates. AI that runs whole pieces of work end to end, inside rules your people set, and brings people the exceptions and anything that needs judgment.

The stages are how you do the work, and the frontiers are where it pays off, and you can move through the stages way faster than we could even a year ago.

I got asked recently, "Won't whatever we build be outdated in a year?" The interface is actually the easy part now. The hard part is getting your business rules/processes/workflows written down and the guardrails set. Once you have that, it doesn't really matter how AI changes.. you take it with you.

5. Who we're for

I've met with over 100 mid-market companies over the past 2+ years, and that's sharpened our picture of who we help. We work with privately held mid-market companies, roughly $30M to $500M in revenue, and the PE firms that own companies like them. Leadership teams where the CEO or COO can make a call in the room, and who want a partner that shows up in person (we're based in Charlotte).

That's where I see this landing hardest. Owners and operators feel disruption coming first, and PE firms keep asking how to help their portfolio companies get ahead of it.

Wrapping up

Getting faster with AI is worth doing, and it's quickly becoming table stakes. Do it, and use it to build your foundation. Just don't let it be the whole plan. Someone in your industry is going to use AI to change how the work gets done, sold or delivered, and it might as well be you. Be the disruptor, not the disrupted.. put a few of your best people on it with a deadline, and start this month.

What's new at StealthX

The new stealthx.co homepage clip, cutting between the Streamline, Stand out and Disrupt labels.

Today we launched a new stealthx.co built around everything above. How we work lays out the 4 stages and 3 frontiers as a TV program guide (yes, we went full retro.. the whole site looks like late-night TV and a shelf of mixtapes now). There's a companion post if you want the framework in 4 minutes.

The new stealthx.co homepage on a laptop and a phone.
The new How we work page on a laptop and a phone.

This newsletter has a new name too: Mid-Market AI. Same newsletter, same me just more specifically focusing on AI for mid-market companies.

Also, v2.0 of the 2027 AI Strategy Playbook is out too. It now has the 4 stages and 3 frontiers built in, plus a page on how to tell which stage you're in. A few things from it worth taking into your 2027 planning:

The cover of the 2027 AI Strategy Playbook 2.0 next to two inside pages.
  • You can be a level 5 on the single player ladder and a level 1 multiplayer. One person can get great with AI on their own while the rest of the team gets none of it. Shared context has to come before a shared AI teammate, or the teammate has nothing to work from.

  • Plan for four cost lines. Seats, consumption (pay per use), foundation and change. Foundation is the one that gets cut because it isn't exciting, but it's the most critical part.

  • The real risk in 2027 is buying AI out of fear of falling behind. A purchase made just to keep up rarely comes with a plan, and the licenses go unused.

It also has a half-day leadership session kit, a one-page AI plan and a 30/60/90 checklist. It's free.. grab it here.

Putting this into action

  • Put your last 3 AI projects on the grid. Which frontier did each one serve? If all 3 are Streamline, you know where the gap is.

  • Name the one change in your industry you'd least want a competitor to make first. Put it on next month's leadership agenda with 2-3 names next to it.

What I'm reading

  • Welcome to the B2A Economy. Olivia and Justine Moore at a16z on the Business to Agent economy, where AI agents shop and buy on behalf of people. "Companies spent decades keeping bots out. Now they'll compete for their business." Great take on something we all need to be paying attention to.

  • Fable 5.2 and Jagged AGI. Ethan Mollick (Wharton professor) got early access to Anthropic's new model and put it through some wild tests. Watch the first video.. it's a couple minutes long and really crazy what it produced from a single prompt. It still overcomplicates little things in ways that show it's not quite at the level of human thinking/creativity.

  • Tolan. An AI guide app for your phone, and one of the most delightful/fun AI agent experiences I've encountered so far.

Onward & upward 🤘

Drew