How to Get the Most Out of your Business Systems Process Mapper Output

Whoa! We’re nearly to the end of October, which means our very own Laura DeVeiga’s favorite holiday is fast approaching, the costumes, are coming out, the candy corn is corning (?), and all of us are getting excited about… process improvement. I know, that took an unexpected turn, didn’t it? But stick with me even if you aren’t; by the end of this you will be too!

A few weeks ago, I shared the Business Systems Process Mapper, an AI prompt that helps you talk through a process in your business and document how it actually works.

A fellow advisor wrote back with a response that made my day. In the past, she struggled to document her processes because the task felt daunting. Using the prompt and talking through the questions made it feel manageable.

Then she asked two excellent questions: How do I learn to write prompts like this? And how do I use AI to improve the process we just documented?

My quick reply was, essentially, “Just ask the AI.”

I stand by that advice, but those questions deserve a fuller answer.

Learning to explain what you want, provide useful context, and guide the conversation will become even more valuable as AI becomes more capable.

You already know something about asking better questions.

Imagine a client tells you, “We want to go to Italy.”

You could immediately send your favorite off-the-shelf Italy itinerary. But we all know that you’ll be far more successful (and the client a heck of a lot happier) if you ask a few questions first.

What interests them? Who’s traveling? What’s their budget? Do they want to see as much as possible or settle into one place?

Those answers help you make a useful recommendation.

We need to apply that same approach when working with AI. Give it a clear goal, some context, and instructions for how you want to work together.

“Improve my follow-up process” is a start and you’ll make progress, but something like “Help me make sure every proposal gets followed up on, using the tools I already have, without creating an hour of extra work each day,” gives it much more direction.

And adding “Ask me questions before recommending changes” gives it a way to uncover what you haven’t explained yet.

Give it a job, a goal, some boundaries, and a finish.

When you want AI to help you with a piece of work, explaining these four things are essential for great results:

• The Job: What do you want help doing?
• The Goal: What would a useful result accomplish?
• The Boundaries: What should it preserve, avoid, or work within?
• The Finish: What should it produce for you?

Then tell it how to conduct the conversation: “Ask two or three questions at a time and wait for my answers.”

You can also ask AI to help write those instructions.

Try this:

“Help me create a reusable prompt for this task. Interview me about what I want to accomplish, then draft the prompt and explain why you structured it that way.”

Read what it produces. Notice how it turns your explanation into instructions. Try it, correct what doesn’t work, and save the version that does. Doing this helps give you practical outputs, while learning how to get from your AI tools at the same time.

Now, let’s improve that process.

The Business Systems Process Mapper outputs a set of information on a given process and was deliberately instructed to document your current process without changing it. But now we actually want to improve the process, so we need a new chat conversation with a different and specific purpose.

Start a new conversation (with fresh context) and paste in the process summary you created with the mapper. Include the flowchart if you have it.

Here’s a prompt you can copy and use (and re-use for other processes):

Help me improve one documented process in my travel advisory business.

I will provide my current process below. First, read it and briefly confirm your understanding. Do not assume missing details.

Interview me about the result I want, where the process causes problems, what already works well, and the time, tools, budget, and help available to me. Ask only two or three questions at a time, then wait for my answers.

Once you understand my situation, suggest up to three practical improvements. For each, explain the problem it addresses, the likely benefit, the effort required, and any tradeoffs. Treat benefits as estimates to test.

Start with changes that use my existing tools. If you suggest automation, identify what we need to verify before relying on it.

Recommend one improvement to try first, explain why, and wait for my decision.

After we agree, document the revised process. Include its trigger, numbered steps, who does what, tools used, important decisions and exceptions, and a clear definition of done. Identify what changed from the original. Create an updated flowchart if useful.

Finally, help me plan a small trial and choose one simple measure to judge whether the change helps. Keep the original process separate from the proposed version.

Here is my current process:

[Paste your process summary and flowchart here.]

As you work through it, it is important to push back when something doesn’t fit. AI systems are powerful, but they make mistakes and are prone to making assumptions, so pushing back from time to time is critical. You can say things like: “That adds too much work.” “My system can’t do that.” “We need to keep that personal phone call.”

Use your judgment and knowledge of your business (and your own strengths and weaknesses) to help shape the output and improve the result.

A better document is only the beginning.

Let’s remember the point of all of this. Hint: It isn’t the document. We’re after continuous improvement, not perfection. Your aim with this should be to find some material improvement in your process, but testing the improvements and adjusting are almost certainly going to be required. So an updated process doc is great, but it’s worthless if you don’t implement it.

So, take what comes out and test it in the real world. Note what worked and what didn’t. Then during your next Architect’s Hour, check whether it helped. Come back to that AI session and share what you observed and refine the process as necessary (Tip: rename each AI session something like “Process Mapper: [process name]” to make finding them easier).

You don’t need to perfect your whole business this week. Choose one recurring frustration, work through it, and put one useful improvement into practice.