No One has Asked for an AI PM (Except Me! - How I AI 4 of 4, Andrea’s Version)

My 2019 AI project manager pitch, what AI still can't replace, and a test: spot what I wrote vs. what Claude did.

How I AI

Andrea’s version · a four-part series

 
 
 
 
Part 4 of 4  ·  November 2026

After all of my tinkering with it, the surprising conclusion (at least to me!) is that AI is making me more grateful for AJC's customer-intimate delivery model even as we use the technology to make us better.

Where real companies actually are:

We’re in the tinkering and "early adoption" stage. The value right now is personal and practical, not transformational, and the big-company data says the same thing. Consider the Chasm (à la Geoffrey Moore) not yet crossed.

If you feel behind, you’re not

56%

of CEOs said AI had produced neither higher revenue nor lower costs over the prior year.

Source: PwC Global CEO Survey, January 2026

73%

of mid-market manufacturers are still testing AI. None had reached full deployment.

Source: Kaufman Rossin, State of AI in the Mid-Market, July 2026

Almost nobody is “done” with AI. Most companies are where we are: seeking the handful of places it genuinely helps.

What’s changed inside AJC

Our team uses Fathom every day for note taking. This area does feel pretty mainstream; I'm often met by an AI notetaker before a human joins my meetings, and you probably are too.

The practical value from Fathom is that when transcripts are combined with our enterprise version of Claude, we can take complex conversations and transition them to immediate draft deliverables, allowing us to deliver a higher level of performance and greater volume of results while maintaining our part-time model. Essentially, everyone wins.

I cannot underscore enough, however, that "draft deliverables" are NOT polished and finished work. Here's a test. Go through this post and see if you can pick out what I wrote by hand versus what Claude wrote "in my voice" and on my behalf. I don't care if you *think* you write like me, Claude/Chat, you really don't, and I can tell.  However, for "Technical" deliverables like bullets or spreadsheet analyses, Claude does just fine.

What hasn’t changed

This is the interesting part. Originally Claude suggested I say: "Nobody has asked us for an AI-only project manager. Not once."

That's actually not true. But the reason it is not true is because I have asked for an AI Project Manager. As in me, Andrea, personally. 

In fact, in 2019 I came up with a product idea called "Hey Jonah" that was going to BE a voice-enabled AI Project Manager that would handle logistics, take notes in meetings, translate them to to-do break-down tasks with due dates, assign them to the appropriate role humans on projects, create project documentation, and even update the myriad software trackers people hate clicking around resulting in always outdated project and ROI metrics.

I even pitched a couple of VCs and to a Pitch Competition, but the feedback in 2019 was that it was "too early," especially the voice-enabled technology. Of course, we did not have the ubiquitous LLM models then that we have now, and today this application is already possible, if only stitched together by early adopters and techies.  

What has not changed, though, is what humans actually are good at doing, and what people won't do for a robot - and that is "change management." You can call it whatever you want, but what I mean by "change management" is the interpersonal connection, individual motivation, and personal satisfaction we get from actual contact with other human beings that make us feel like what we are doing is seen, matters, and adds real value to the greater good of the group as a whole. That group could be a company, a department, a team, or even a duo, and is NOT replaceable by artificial connection, no matter how intelligent or complimentary it may be.

AI will enable humans to be more productive, we will get more technical work done, and there are no excuses for humans to not seek out information and create tangible results using AI, because it is easier to get at these assets than ever. But we want our work to matter to SOMEONE. We want our SELVES to matter to someone. 

So AJC's core competitive advantage is even more potent now than ever before, IMHO, because we've always won on "customer intimacy." We become a part of your team for the time we work with you, and honestly, we care about our customers for a lifetime. It's personal to us, because it is personal to you. And we are better at the work itself now than ever before thanks to AI.  

 

What companies should pay humans for:

Connection. Knowing the people well enough to know what they’re not saying.

Critical thinking on execution. Seeing that step six will fail before anyone starts step one.

Discernment about pressure. Knowing when a project competing with someone’s day job needs more push, and when it needs more grace.

Focus. Keeping good people from getting pulled back into the urgent and away from the important.

AI Adoption is still a human problem

Gallup published in September 2026 that workers who use AI often are about twice as likely to fear losing their jobs to it. But when workers have a manager who actively supports their use of AI, that fear drops substantially.

That’s change management, and it’s the same for any new thing, whether it is AI transformation, an ERP/systems infrastructure backbone, or a business model shift. The technology can be "perfect" but whether anyone actually adopts it so your company sees the ROI, still comes down to people. If no one cares enough to make it work for me, why should I care enough to make more money for you? 

Fair point.

Three guesses about 2027

The novelty wears off; judgment gets more valuable. When everyone can produce a polished document in thirty seconds, the scarce thing is knowing which document matters. [AI wrote that, and I 100% endorse it; no excuses for not being up-to-speed on the polished docs and exercising human judgement using the info.]

Your data decides your AI. Most of the mid-market companies I talk to don’t have an AI problem. They have a messy-systems problem that AI makes more obvious. The ERP decision you’ve been putting off is now also your AI decision. [AI wrote that too, and again I 100% endorse. Hard to have a digital worker when your environment is not digital.]

The best uses will be boring *and not obvious ROI drivers*. Notes, follow-ups, first drafts, and cleanup, freeing people up for the hard conversations only they can have. [*My add - because AI will be a tool. Who points to Microsoft Excel and says "My ROI from using Excel over the last 25 years is XYZ." I mean, you can guesstimate, but humans would find a way to make big move-the-needle plays like M&A, NPI, key accounts, etc. happen Excel or no Excel. The Excel TOOL made it a lot easier for the humans taking the action, and that's worth something. Fuzzy on the down-to-the-penny ROI as to how much.] 

An update on my AI clone

Since Part 3 went out, I gave my Delphi a new job: front desk for intro-call requests. I get a lot of “Can I pick your brain?” asks, and I want every one of them to get a fair shake. But my time is finite, and a lot of those asks turn out to be something else: a sales pitch, an introduction, an investment, or a job request.

So I wrote my Delphi a screening manual. Before anyone gets time on my calendar, my clone helps them sharpen the ask: what they actually want, why me, and what’s in it for both of us. If it can answer from what I’ve already written, it does. If it can’t, it isn’t allowed to guess. It writes me a one-page brief instead.

Game-changer or awkward robot receptionist? We’ll see. It’s a pilot. If you want to help me test it, try asking my Delphi for something, and I’ll let you know how it goes.

What you told me

People wanted me to add their colleagues to this newsletter series, I got responses from some folks whom I didn't even realize received it, when I posted it on LinkedIn, some people signed up from that promo channel, and I'm hopeful there were a few forwards. If this was a valuable series and you'd like more like it, drop me a line. 

Who is willing to let me interview you to share with others how YOU use AI in your company? 

Lots of people say they hope their experience will help others. I'd love to be a vehicle for that to happen for you. 

Book a time with Andrea →

Thanks for reading all four. Writing this series (mostly) by hand reminded me why I like doing it that way.*

*I think it is hilarious that other than the (mostly), Claude wrote that last sentence. That is so AI. Ha!

Talk soon,

Andrea

Andrea Jones · Principal, AJC Company

Andrea Jones

Founder. Innovator. Speaker. Consultant.

https://ajccompany.com
Next
Next

I Cloned Myself. It Hasn’t Made a Dime.(How I AI 3 of 4, Andrea’s Version)