Read on: My website
Read time: 2 Minutes
You do not have an AI problem.
You have an honesty and AI discipline problem.
You keep getting told the same story: your firm is “early,” the models are “not mature enough for us,” the value is “right around the corner,” or the AI tools need your company to “help them pilot for the market.”
The list goes on, and the specific tools might not be ready for your individual needs today.
Meanwhile, you have spent real money, burned real time, and you still cannot point to a single AI project your board would call a win.
Was this newsletter forwarded to you?
You are not running one AI problem; you are running two completely different ones, and mashing them together is exactly why your ROI keeps vanishing, your frustration keeps getting higher, and blood is most likely at a boiling point.
And you probably don’t want to keep laying your credit card down for anything new.
This week’s newsletter is about separating those two problems, fixing the part that is quietly taxing your best people, and building the part that will actually move your margins, not just your slideware.
Let’s dance.
Reply to everything. Edit nothing.
Your inbox is full. Slack is piling up. Client messages need a response yesterday. Typing thoughtful replies to all of it takes hours you don’t have.
Wispr Flow turns your voice into clean, professional text you can send the moment you stop talking. Speak like you would to a colleague — tangents and all — and get polished output. Emails, Slack, LinkedIn, WhatsApp, whatever’s open.
89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Works on Mac, Windows, and iPhone.
First Touch – Share Something Useful
Under the hood, your business is doing two very different jobs with AI.
The first is service delivery. That is all the work your clients see and pay for. Strategy, analysis, implementation, advisory, roadmaps, recommendations, proposal development, the list goes on.
When you use AI to gather requirements faster, analyze a pile of research, construct a sales pitch, or draft a stronger first version of a deck, that is delivery AI in action. The work is probabilistic. It is judgment-heavy. A human is always going to make the final call and add (hopefully) a little customer flavor, edits around the market, the opportunity, or the specific proposal.
AI just helps them get you there faster.
The second is services management. That is everything that keeps the machine running. Staffing. Project Management, check-in calls, Project margins. Billing. Revenue recognition. Renewals. The internal plumbing that your clients do not see but instantly feel is about to break.
This work is deterministic. There is no “close enough” when it comes to revenue schedules, margin calculations, or invoices. If it is wrong, it is wrong.
Those two domains are not the same. They do not tolerate the same kind of error. They do not need the same kind of engagement, prep, delivery, or follow-up. How you leverage AI across these two domains must be tailored, as you cannot rinse and repeat what you used for one for the other.
You can probably already see where that goes wrong.
Want to check out some of my other recent articles here:
The Hidden Cost Of AI: Skill Erosion
The Importance of Executive-Level Sponsorship
3 Simple AI Prompts Tricks You Should Use Daily
Verification Tax
We want to position how AI can work for you in delivery.
Ask a team member to pull together a point of view on a new market. How do they approach it (nowadays)? For the sake of this exercise, everyone is different; some go off of data, from their years in the industry, who they know, etc.
How will they do it, working with AI (regardless of their age and years in the industry)?
Instead of starting from a blank page, they would probably drop the question and a few prompts into an AI tool, get a decent synthesis of twenty articles, and build their own recommendation on top.
They might rewrite key parts. If they have some industry insights, they would be able to provide that feedback back with AI to provide additional enhancement. They still run it through their judgment. But the start is faster and better than it used to be. That is a win.
Now think about how AI is “working” for you in this specific domain.
A vendor plugs a generative AI layer into your project systems. It spits out “smart” margin projections, draft billing schedules, and suggested revenue plans. The demo looks slick. The moment you try to use it for anything that actually matters, your senior project managers and finance leaders do not trust it.
So what do they do?
-
They review every output
-
They cross-check every number
-
They reconcile the AI’s view with the system of record
You have not automated the work. You have just turned your most expensive people into full-time AI auditors.
That is the verification tax.
And it can add up…quickly.
The hidden cost of asking a probabilistic system to run deterministic workflows. In delivery, human review is expected and valuable.
In management, mandatory human review means the supposed “automation” failed. The savings you thought you were getting are being erased by the headcount you are using to clean it up.
You feel the friction. The skepticism. The “This seems neat, but I still have to check everything.” That feeling is your ROI draining right in front of you.
Zoom Out
When you zoom out, your experience is not unique.
A 2025 MIT-linked report found that roughly 95 percent of generative AI pilots failed to deliver measurable bottom-line impact. McKinsey’s 2025 State of AI work says only about 6 percent of organizations qualify as true high performers, meaning they can point to significant value and at least 5 percent of EBIT tied to AI.
You are living in the middle 94 percent. Lots of experimentation. Not a lot of earnings.
This is not because AI “does not work.” It is because of how it has been introduced.
You rolled AI out as a horizontal layer. A copilot here. A chat window there. A plugin that sits next to your systems instead of inside them. You treated it like one generic capability, not as two very different tools for two very different jobs.
On top of that, your data foundation is probably weaker than you want to admit. Industry research has been saying the same thing for two years: data quality is the single biggest barrier to real AI adoption. When finance, delivery, and customer systems do not share a unified data model, any “AI for operations” is guessing across gaps. That is why you stay stuck in document-adjacent use cases instead of transforming the workflows where the real money lives.
So yes, you are in pilot purgatory. But there is a way out.
Start treating your two AI problems like they are actually different.
Separate the Two
On the services delivery side, your job is not to replace your team. Your job is to arm them.
You should be asking:
-
Where can AI cut research time in half?
-
Where can it produce higher-quality first drafts?
-
Where can it help juniors think more like seniors, faster?
If the tool gets you 70 or 80 percent of the way there, that is a win, not a risk.
On the services management side, the question is totally different.
You should be asking:
-
Which workflows can be run by deterministic, domain-specific AI?
-
Where can agents actually execute, not just suggest?
-
How do you eliminate manual checking, instead of just moving it around?
This is operations. If your platform has deep, integrated access to staffing data, financial rules, project status, contract terms, and risk thresholds, how are you leveraging AI? Don’t sugar something on top.
McKinsey’s work shows that high performers are nearly three times as likely as everyone else to have rebuilt workflows around AI, rather than simply layering AI onto existing processes. That is what this looks like in practice. Fewer “assistants” that sit on the side. More systems that quietly do real work in the background, with clear control points and alerts where you actually need humans.
Now take one more step.
If you reclaim 20 hours a month for each project manager by automating staffing, margin checks, and billing administration, that is not a nice productivity story. Now run that across your delivery organization; it is a strategic shift in how you deploy your most expensive people. Less time chasing hours and invoices. More time solving client problems and growing accounts.
Where to Go
The next wave is not “better chat.” It is agents that can run real, cross-functional workflows without you hovering over every decision.
To get there, you need both sides of your house to talk to each other.
A staffing agent cannot just see who is free next week. It needs to know what skills the project actually demands, how far over or under scope you are, what margin you are targeting, what the contract allows, and what the relationship dynamics look like. Some of that is delivery data. Some of that is management data.
Likewise, an AI that helps a client-facing team decide what to do next cannot just see “best practice.” It needs to understand actual capacity, financial constraints, and risk appetite, all of which live in your operational layer.
The key is this: you build the two sides differently today, but you build them toward each other.
This is all how you want to approach this.
Are you going to keep saying “AI just needs more time,” or are you willing to admit you have been solving one big, vague AI problem instead of breaking it into two very clear ones?
If you keep treating AI like a generic layer, you will keep paying the verification tax, keep piling up pilots, and keep explaining to your board why the slides look better than the numbers.
If you separate delivery from management, design for each on purpose, and start building the data and systems that connect them, you give yourself a different outcome. AI that actually makes your people better. AI that actually runs your operations. A platform that can support real agents when you are ready.
You can do the same.
See you next week.
Whenever You’re Ready, Here are 4 Ways I Can Help You:
-
Unlocking Hidden Potential – Reconnecting with Past Clients for Explosive Growth – Check out my free eBook on how you can find hidden gems in your past clients and help you crush your sales goals.
-
AI for Business Development – Download our free eBook on how you can effectively leverage AI prompts to your advantage. From properly setting up your preferred AI tool, to how to shape your prompts, save time, and get the outputs you are looking for.
-
Sales Resources at Your Fingertips – From tools, tips, demos, and how-tos, check out our Pages and content that can provide you with additional support, whether it be social selling, account management, or something else.
-
Cribworks Advisor Program – Want more than just resources? Reach out to me and see if our Advisor Program can help you scale your business.
