Last week, I made the case for why the wait-and-see strategy that worked for various technology upgrades like ERP and CRM won't work for AI. The tools commoditize either way. What compounds is the data and the institutional knowledge you build on top of them, and that window is open right now in a way it never has been before.

I want to be precise here, because “AI is going to change everything” is not an argument. It's noise that isn't helpful. That said, AI will most certainly transform the way we work in the future. For the entire history of your company, the constraint on commercial performance has been some combination of capital, equipment, labor, and market access. You could only run so many machines. You could only hire so many people. You could only reach so many customers with so many salespeople. Every commercial growth problem was ultimately a resource allocation problem.

That constraint hasn't gone away. But AI enables us to unlock human efficiency and intelligence at scale.

Think about what it would mean to have every deal in your pipeline flagged for risk in real time? AI can find those risks based on pattern analysis instead of gut feel, so you have the ability to find out the moment an opp starts going cold, instead of finding out on a forecast call three weeks later. What if you could have account and territory performance rolling up continuously, instead of a quarterly fire drill to build the board deck. OR, what if you could know (without months of analytical work) which marketing and sales activities drove the deals that closed this quarter, instead of guessing at it in the QBR? I'll take it one step further: what if that intelligence happened in a way to automatically adjusted your marketing and sales strategies moving forward (the ultimate continuous improvement loop!)

None of this requires new customers, new products, or new headcount (and if you already have an LLM subscription, it doesn't even require any new technology). It requires applying intelligence, consistently and without fatigue, to the commercial data you're already generating and mostly not using.

I find it useful to think about where industrial companies sit right now relative to three distinct eras of capability.

The industrial revolution built your core competency: physical optimization. You know how to move materials efficiently, run machines reliably, and eliminate waste on the plant floor. Most of the industrial companies I work with are genuinely world-class at this. The plant runs “lean”. OEE is measured. Kaizen events happen. Standard work is documented. 5S occurs regularly. That's real, earned capability that took decades to build.

The digital revolution asked something harder: information optimization. Capture the data, move it across systems, use it to make better decisions. Most industrial companies built the systems but never really finished the job. The ERP has the history, but nobody queries it systematically. The CRM exists, but nobody fully trusts the pipeline data in it. The data is there, fragmented, inconsistent, and largely inaccessible without someone manually pulling it. Information optimization was attempted. It was never completed.

For a while, I thought the fix was to finish that job first: get the process fully documented, get the data fully clean, and only then bring AI in on top of it. I don't think that anymore. The reason we couldn't see the full ROI out of the digital revolution has less to do with the SOP than with who has to keep it living after it's written. Until now, the humans doing the work have also been responsible for following the standard work. Someone still must keep the CRM current, pull the data out, turn it into an actual decision, and hold the team accountable to running the process the way it's written down. Further, most of these teams don't have enough headcount to do the core commercial work in front of them, let alone spare someone to hold the process and the data behind it accountable. The rigor required to maintain process discipline in commercial contexts is superhuman.

AI finishes the digital revolution. When you build an AI step directly into how a rep logs a call, or how a quote gets entered, the process gets documented and the data gets clean as a byproduct of people using the thing. And AI has the ability to make doing much of the work of our jobs faster and easier… so when the thing that you hate doing about your job becomes automatic AND reinforces the process, it is magical. The foundation gets built by doing the work itself, in real time, instead of sitting in a separate phase everyone has to remember to get to. That's the same sequencing mistake most companies are about to make with AI that they already made once with their CRM: treating the foundation as a separate phase instead of building it into the work itself.

The good news is that it's the same discipline that made your plant world-class, just applied in a different order. You already know how to do this. You just haven't applied it to your commercial function, yet.

Once you see it that way, the next question is who does the building, and this is where a lot of companies are defaulting to the wrong answer.

Every person doing a job in your commercial function is sitting on use cases for AI that nobody else can see. There are hundreds of them buried inside a single team, and the only people positioned to find them are the people doing the work every day. AI cannot be applied to a function by someone outside of that function … not an outside consultant, not a software vendor promising to bolt AI onto what you already have. AI, underneath all of it, is a simple set of building blocks and tools. You don't need to be a “technical” person, you simply have to be able to communicate and understand how to think like the AI tool. Put together well, those blocks make teams and individuals faster. They clear out the busywork, help prioritize what matters, and free people up for the work that needs a person: setting vision, thinking strategically, etc.

No outside party and no centralized function inside your business is ever going to show you a real return on AI across an industrial organization. That return shows up when every person on the team is using AI well. The role of the organization in all of this is governance and coordination. Can you imagine if we only let one department have computers, and we said to everyone else: You come to us, and we will do the computer things for the business? Yeah, no way. AI is the same way. AI isn't something that lives just within the IT function. It truly has to be owned and embraced by everyone across the organization.

Go back to the person and the answer from last week. Remember what they said when you asked what feels like busywork? Don't write a process document about it first. Instead, tell your LLM (Claude, ChatGPT) about the busy work that keeps repeating, and ask it to help you solve it. Watch what happens to your, or their, excitement about AI once you see what it can help you solve for real, instead of hypothetically.

To get you started, here is your prompt you can use (just copy and paste it into the chat window in your preferred LLM, like Claude or ChatGPT):

That's the foundation. It's the highest-leverage thing you can do before you spend a dollar on anything else.

Part 3 closes the series: concrete examples of what this looks like, already running inside companies like yours, and why learning to use AI is a series of small, sequential building blocks, not one big leap.