Business Process Automation: What It Actually Takes to Do It Right

A business has to have processes. So business process automation is the natural next step — not some exotic upgrade, just the logical thing you do once you realize you're running the same plays over and over by hand.

And here's the part that should encourage you: if you have employees, and you've ever laid out a job for one of them, you've already done about eighty percent of the work it takes to build an automated process. You've thought about what needs to happen, in what order, by whom. That instinct is the foundation. The rest is discipline.

But that last twenty percent is where almost everyone underestimates the job. So let me walk you through what it actually takes — from the chair of someone who hunts these processes for a living, not from a software brochure.

An automation has to know everything an employee never had to be told

Most business owners leave a lot to the employee. It's natural. You'll often work alongside someone for years before you ever fully hand them a responsibility, and by then they've absorbed a thousand small judgments you never had to spell out. They know what to do when the form comes in wrong. They know who to call when something's off. They fill the gaps without being asked.

An automation gets none of that for free. It needs to know everything. It doesn't stop at the general idea of the task — it needs every step, every exception, every "well, if that happens, then we do this instead." And the moment you add AI to it, you have to build more than you think, because now you're not just scripting steps, you're teaching judgment.

There's a layer to this I call the software layer. It's all the if-this-then-that — the pathways that keep the system running when something unexpected shows up. The good news is that most of this already exists. It's in your head. You've just never had to say it out loud, because you've never had to explain it to something that couldn't infer it. Drawing it out, if you've got someone who does this for a living, is straightforward. Drawing it out of your own head, alone, can feel like an impossible task.

Picture a guy who's run a heavy piece of equipment for twenty years. He's been doing it so long he probably couldn't fully tell you what he's doing. He's got moves, techniques, little corrections he makes without thinking — but he doesn't see them as techniques that could be written down or taught. To him, that's just what you do. The expertise is invisible to the person who has it. Pulling it into the open is the actual work of automation, and it's why this can't be done from a brochure.

There's a process to finding the process

There's a process to everything, and building one for automation is no different. We start with an interview with the owner. Then we talk to anyone we can in the business. Everyone has a voice, because everyone is involved — and the person doing the task often knows things the owner doesn't.

The most common thing I run into is that people zoom in too close to the problem they're feeling. So close that they don't realize the problem can't actually be solved at the spot where it hurts.

Here's a small example that shows why pulling back matters. Say your supplier faxes you an inventory list. You hand that paper to your operations manager. He goes out and checks everything that came off the truck against the list. Your complaint is that things keep getting missed — and despite his best effort, sometimes they are, and it costs you money. It's easy to look at that and decide the problem is the operations manager.

It isn't. The problem is upstream. A supplier faxing you a piece of paper is not built for the digital age. If that same list came as an email or a file — a simple .csv — it could feed a scanner that reads the barcode of every item and checks each one off automatically. Nothing slips through, because a machine doesn't get tired or distracted on item forty of sixty. The fix was never at the point of pain. It was a step back from it. That's why we look everywhere, not just where it hurts.

And there's a bonus to looking everywhere. Sometimes you come in focused on one issue, and by the time we're done we've found more savings and more value in something completely unrelated — a leak nobody was even looking at. The value is in the digging and the collection of data. That's step one, every time.

The story that taught a whole agency what it actually does

A while back a home health care agency came to me. They wanted something to help screen applicants and collect their information more efficiently. Reasonable ask. I expected a software conversation.

Then I walked into their offices and found shelves and shelves of paper files. So I did what I always do — I started talking to each person about how someone actually applies to work there. And I was genuinely shocked: nobody knew the whole process. It was scattered in bits and pieces across several people, every one of them pointing to someone else who might know the rest. There was no single human being who held the entire thing.

So I had my team build out a basic application process we could grow on — something concrete I could put in front of them to surface the missing pieces I could see were there. For the next two months we pushed real applicants through it, and management would come back with complaints about things missing. They didn't understand what I was asking for. In their minds, things just... happened.

Take one piece of it. I asked, "Does an applicant need to upload a TB test?" Yes. Anything special about it? No, of course not — it's just a paper. Does it have an expiration date? Oh, TB tests don't expire. So they never have to replace it? Oh, yes they do — at the end of the year, one year from when they took the test. Oh. So it does expire. Good to know.

Then about four weeks in, the manager comes to me and says these are chest x-rays. I'm a computer guy — I'm confused. He's right, though: people had been uploading chest x-rays as their TB test. Why would they do that? Because a chest x-ray is good for five years, he says.

Oh. That's good to know, too.

I'd assumed he'd have handed me all of that up front. But he only knew how to handle it — he'd never had to explain it. Because I didn't have the right question, he didn't know how to give me the answer I needed. It wasn't hiding. It was just invisible, the same way the equipment operator's techniques are invisible. Once we finally got it out into the open, that agency will never miss it again. That's what the discipline buys you.

The discipline is the whole thing

Everyone naturally looks for the easy way out. With business process automation, you might not be looking for the easy way — and here's why.

To build one of these, you have to collect a mountain of information and look at every single movement. The work is tedious. At points it feels almost ridiculous, like you're documenting things too obvious to write down. But that diligence is exactly what makes it work, because when it's finished, it runs the same way every time, over and over, without drift. It's honestly a beautiful thing to watch. And without that completeness — without being willing to make it complicated where it needs to be complicated — it can never become what it's supposed to be.

The discipline is to map out everything: all the paths, all the spaces in between, all the contingencies. That's what makes a process smart. And with AI on top, we can make it smarter still. AI agents can make genuinely good logical choices — but only if they know how a manager would handle every situation. Which sends you right back to the interviews, mapping out case after case after case. Put in that work, and the end result is clean.

You can over-automate, too

Here's something fascinating about this work: sometimes you can automate too much. There's a thing we call the human element, and it's real — some moments need a person in the loop, not because the machine can't act, but because a person needs to know it acted.

Simple example. It might be easy to fire an email off to a client automatically, quietly, in the background. But the person managing that client relationship needs to know it went out. So the process might deliberately pause before sending — the manager clicks "yes," and now they're informed and in control. We add points of reference like that on purpose, to keep a human in the loop on what's happening. Sometimes we build visuals, so the person running the system can look up and actually see the state of things at a glance.

Automation done right isn't about removing every human. It's about removing the tedium and keeping the human exactly where their awareness and judgment matter. (We made the same point about customer service in how to automate customer service without driving customers away — automate the routine, keep the person for the moments that need one.)

Why the off-the-shelf tool so often disappoints

Tools can be really appealing. They look like they'll solve all your problems, and maybe, in some narrow way, they will. But more often than not, they don't add up to what they promised. They bring connection problems, things that won't talk to each other properly, and you end up with one more thing to maintain.

And here's how it actually fails. You end up with another unmaintained trigger point that quietly gets abandoned. What was supposed to be automatic turned out to have one new button somebody had to press — and one day nobody presses it. Then nobody presses it the next day either. Soon the whole thing isn't really working, because the part that was supposed to run on its own had a human dependency baked in that no one accounted for. The sales agent swore it was no big deal. It ends up being a big deal.

That's the difference between a stitched-together tool and a process built to actually run. (It's the same argument we made about business process automation tools — the tool was never the point; the workflow is.)

"Will this replace people?"

This is the question that always comes up, and I always take a step back before answering it.

If everyone tried to protect their own job, innovation would have stalled out a century ago. The horse-and-carriage driver. The factory worker. Progress has never waited for people to feel safe about it. So when it comes down to jobs, my honest answer is this: if you're useful, there's a place for you.

Think about what COVID did. A lot of people got a once-in-a-lifetime chance to show their bosses what it would look like to let them work from home. And sadly, a lot of them proved the opposite of what they intended — they showed that, for them, working from home was a bad idea. But not everyone. I know someone who pushed his production to a whole new level the moment he got the chance to work from home. So when everyone else was being called back in, he got to stay where he was, doing what he loved — because he'd made himself undeniably valuable.

That's the whole point. The reality is that if a position can be replaced by AI, it's because the position was inefficient to begin with. Your employees — all of us, honestly — should be pushing to be productive and valuable. I focus on being valuable, and if my position gets wiped out tomorrow, I'll go on being valuable somewhere else. There is always room for valuable people. Automation doesn't change that. It just raises the bar on what valuable means.

You can wait. For a while.

So what happens if a business owner just keeps going as they are and never automates anything?

Honestly? For a while, it's not a big deal. You can coast on the way things have always worked longer than you'd think.

Publix took its time adding self-checkout, and their reasoning was the personable experience — they didn't want to lose the human touch. Fair enough. But as customers everywhere got used to those machines and started to enjoy the convenience, holding out stopped looking principled and started looking behind the times. They had to meet customers where their expectations had moved.

It's the same lesson Amazon has been teaching everyone else the hard way. Sub-24-hour shipping and no-hassle returns sounded like an overreach when they started chasing it, and plenty of retailers treated it like a bluff. Those same companies are scrambling now. Customer expectations only ratchet one direction.

So yes — you can dodge automation. But eventually your clients get annoyed that your service isn't as fast, as smooth, or as reliable as the business down the street that didn't wait. Growing isn't optional forever. Whether you feel fine right now or you're already a little worried, improvement is something you should be doing — and business process automation is a great place to start.

When you're ready to find out what's actually worth automating in your business, that's exactly what we do in an Assessment. We come in, we talk to everyone, we map how the work really flows, and we find where the money is leaking — including the leaks you didn't come to us about. The tools were never the point. The process always was.