Key Takeaway:
Understand why your expertise, not AI, must be the starting point for any process improvement effort
Learn a three-part prompting method that directs AI to review your SOPs for clarity and hidden gaps
Identify the “assumption gap”, the most common reason a process breaks down when someone new follows it
Gain a repeatable approach to using AI as a critical thinking partner, not a shortcut
You documented your process. You created the SOP. You followed all the advice about transferring your systems from your brain onto paper.
So, why does something still seem a little off?
More often than not, the issue isn’t the documentation but the review process. Most small business owners write a process once and then move on. However, a process you’ve written yourself is also something you’re too close to. You know every step so well that you stop noticing the gaps, assumptions, and points where someone without your experience might quietly get lost. That’s exactly where AI can be helpful, if you use it correctly.
If you have not yet read our earlier post on why SOPs matter for small businesses and how to get started, that is a great place to begin. This post continues where that one left off, specifically focusing on how AI can review, challenge, and improve your existing processes.
You Are the Expert. AI Is the Reviewer.
Here is the mindset shift that changes everything: AI should not be writing your processes for you. It should be reviewing the ones you have already built.
This isn’t a subtle distinction. When you ask AI to develop a process from scratch, it provides a generic framework based on broad knowledge. It doesn’t understand your customers, your team, your tools, or the specific details that make your business operate the way it does. That knowledge belongs to you, and it is often your small business’s uniqueness and approach that set you apart from your competition.
What AI is truly good at is evaluating what you give it. It can read your documented process and ask the questions a fresh set of eyes would ask. It can identify where your instructions assume prior knowledge your team may lack. It can highlight where a step seems complete to you but would leave someone else confused about what to do next.
The process-first approach involves first building the system, then incorporating AI to evaluate and challenge it.
“The best use of AI is to challenge your thinking, not do your thinking for you. Your experience is the asset. AI is the tool that helps you see what you are too close to notice.”
– Jennifer Shaheen
President and Founder, Technology Therapy® Group
The Four-Part Prompt That Makes AI Useful
Vague prompts produce vague feedback. If you paste your SOP into an AI tool and type “what do you think?” or “Review my process and tell me what’s missing”, you will get a response that sounds thorough but does not actually help you improve anything.
Here is the structure that works. Every prompt you use to review a process should include three things:
1. A role
Before you share anything, tell the AI what perspective to take. This frames how it evaluates your work.
- Reviewing a customer service process? “You are a customer service operations specialist.”
- Reviewing an order fulfillment workflow? “You are an operations manager for a small retail business.”
- Reviewing a marketing process? “You are a digital marketing strategist.”
The role is not decoration. It tells AI what standards to hold your process against.
2. The audience
Specify exactly who will follow this process. A procedure designed for a ten-year employee sounds quite different from one for a new hire in their first week. AI needs to know the target audience to accurately evaluate the document.
3. A specific objective
Do not ask for general feedback. Ask for two things specifically:
- Is this process clear to the intended audience?
- Is there anything missing? Is there a step I assumed that someone without my background would not automatically know?
A complete prompt might look like this: “You are a customer service operations specialist. Review this SOP for a retail returns process. The audience is a new part-time employee with no prior retail experience. Tell me: is this process clear enough for that person to follow independently? And is there any step I may have assumed that they would not know?”
That level of specificity is what produces feedback you can actually use.
4. Think, clarify, and explain
Not all AI platforms approach things the same way, and as models evolve, it is important to include clear guidance in your prompt to assist both the platform and yourself.
If you’re not using a recent thinking model, instruct the AI to pause and think before responding. This isn’t needed with Claude’s Opus 4.7 or ChatGPT’s thinking model, but it’s helpful to include because users often forget to verify which model they are using.
Ask the AI to ask you any clarifying questions about the process you have it reviewing:
- • What objectives or outcomes do you see after reviewing my SOP?
- Are there missing threads of information?
- Challenge my process and ask me questions about my approach.
I find this extremely helpful for the AI and for me. This is the challenge you need, especially if you are working on your own. The AI can be a sounding board if you prompt it to be one.
Don’t just accept feedback; ask for the AI’s reasoning or explanation. This last step gives you understanding and often helps you rethink, refine, or rebut the feedback you’re getting from your AI evaluation.
“A good prompt isn’t a question, it’s a brief. Tell AI who it is, who it’s talking to, what you need, and ask it to show its thinking. That’s how you turn a generic answer into useful feedback.”
– Jennifer Shaheen
President and Founder, Technology Therapy® Group
The Assumption Gap: Where Most Processes Break Down
There is a specific kind of missing step that appears in nearly every documented process we examine. We refer to it as the assumption gap.
It is the moment between Step 3 and Step 4; for example, when an experienced person thinks the path is obvious, but someone new gets confused and stops. The expert author skipped this part because they have done it a thousand times. The new reader hits a wall. I have seen this far too many times in tech manual instructions. (Psst: Meta & Google, are you listening?)
This isn’t a failure of effort; it’s a failure of perspective. You’re just too close to your own process to see it clearly.
When you prompt AI to review your SOP, ask this question directly: “Is there a step I assumed between [X] and [Y] that someone without my training might not understand?” Remember, the AI has experience too, so this prompt will push it to think, and it may even prompt you to share your experience for context.
That particular question often yields the most valuable feedback. It reveals the unseen steps that exist in your mind but never appear on the page.
What to Do with the AI Feedback
Getting AI feedback is not the finish line. What you do next determines whether it is useful.
Go back through your process with the AI’s notes in hand and ask yourself honestly: Do I agree with this? Some suggestions will be immediately obvious improvements. Others will miss the mark, usually because you did not give AI enough context.
If a piece of feedback does not make sense, that is a signal to look at your prompt, not necessarily your process. Ask yourself: Did I give AI enough information to understand this step? If the answer is no, add the missing context and run it again.
You are always the evaluator. AI surfaces possibilities. You decide what is valid.
Remember, as I emphasized earlier—and I think it’s worth repeating—ask AI to clarify its reasoning. After receiving its feedback, follow up with:
“Why did you make this recommendation? Where in the document did you find this?”
This prompts AI to base its suggestions on what you actually wrote; especially when using certain AI models, and it helps you quickly distinguish useful insights from generic advice.
“The value isn’t in AI’s answer, it’s in the questions AI prompts you to ask yourself. Every assumption it reveals is a gap you can address before it costs you a customer or a new hire.”
– Jennifer Shaheen
President and Founder, Technology Therapy® Group
From Better Process to Better Systems
Once your processes are solid, a natural next step follows. When a process is clear, complete, and documented, it becomes buildable. You can use it as the foundation for team training, automation, or tools that help your business grow without needing your constant presence.
But that only works if the foundation is solid. A vague or assumption-heavy SOP does not scale. It simply causes confusion at a larger volume.
The sequence matters: clarity first, then systems, then scale.
Your processes represent some of the most valuable operational knowledge your business possesses. They are worth the time dedicated to reviewing them carefully. And with AI as your reviewer, you have a tool that can help you uncover what experience has hidden, as long as you stay in the driver’s seat.
Is Your Process Ready to Scale?
We work with small business owners to evaluate their processes, identify gaps, and build smarter systems with AI. Let’s talk about what that looks like for your business.
