AI Reality Check Worksheet for Your Business

Jump to component

Component 1

AI capability map

Use this map after Lesson 1 to help learners separate different AI use cases.

AI category	What it does	Small business examples	Watch-out
Generative AI	Creates or rewrites text, images, code, audio, or structured drafts	Email drafts, blog outlines, social captions, proposal first drafts, SOP drafts	Can sound confident while being wrong or generic
Machine learning	Finds patterns in data and makes predictions or recommendations	Lead scoring, churn risk, fraud detection, product recommendations, demand forecasting	Needs organized, reliable data to be useful
AI-assisted automation	Runs repeatable workflows with AI steps inside them	Form intake summaries, support ticket routing, CRM updates, follow-up drafts	Needs clear rules, review points, and error handling
AI agents	Uses tools or follows multi-step instructions to complete a larger task	Research assistant, support assistant, internal workflow assistant, reporting assistant	Requires tighter boundaries, permissions, and oversight

Component 2

AI assistance vs. AI replacement

Question	AI assistance	AI replacement
Who makes the final decision?	A human reviews and decides	The system acts with little or no review
Best for	Drafting, organizing, summarizing, researching, preparing	Narrow, low-risk, repeatable actions
Risk level	Lower when review is built in	Higher when customer trust, money, or sensitive data is involved
Good first step?	Yes	Only after the process is tested and low-risk

Rule of thumb: Start with AI assistance. Move toward replacement only when the task is narrow, the risk is low, the rules are clear, and the workflow has been tested.

Component 3

Problem-first AI sentence builder

Use this before signing up for any AI tool.

Template

I want AI to help me [specific task]
so I can [specific business outcome],
using [inputs/context],
while respecting [constraints/risk limits],
with [human review step] before the final output is used.

Build your sentence

Specific task

Specific business outcome

Inputs/context

Constraints/risk limits

Human review step

Your sentence

Example 1: service business

I want AI to help me turn estimate request forms into first-draft follow-up emails
so I can respond to new leads faster,
using the customer's form answers, service area, and project type,
while avoiding price promises or scheduling commitments,
with a team member reviewing every email before it is sent.

Example 2: content workflow

I want AI to help me turn customer questions into weekly newsletter ideas
so I can publish more consistently,
using sales call notes, support emails, and common objections,
while keeping our advice accurate and on-brand,
with final review by the business owner before publication.

Example 3: operations

I want AI to help me summarize weekly project updates
so I can see blockers faster,
using task comments, meeting notes, and deadline changes,
while keeping private client details out of public tools,
with a manager reviewing the summary before it is shared.

Component 4

AI fit quick filter

Use this light filter in Module 1 before the deeper AI Opportunity Matrix in Module 2.

Score each item from 1 to 5.

Task	Repeats often	Uses text/data	Has clear inputs	Needs human judgment	Risk if wrong	Good first AI candidate?
Example: summarize sales call notes	5	5	4	3	2	Yes, with review

How to read it

Strong early candidates repeat often, use text or data, and have clear inputs.
Be careful when human judgment is high or the risk of being wrong is high.
A task can still be a good AI-assisted candidate even when it is not safe to automate fully.

Component 5

Do not use AI yet when…

Use this as an early safety filter.

Do not use AI for a workflow yet when: You cannot clearly explain the task. You do not know what a good output looks like. The inputs contain sensitive customer, employee, financial, legal, or health information and you have not reviewed the tool’s data terms. A wrong answer could harm a customer, create a legal issue, damage trust, or cost the business significant money. Nobody is available to review the output. The process is already broken and AI would only make the broken process run faster. You cannot measure whether the AI workflow is helping.

Better move: simplify the task, document the process, remove sensitive data, add a human review step, then test AI on a small sample.

Component 6

AI reality check worksheet

Use this as the Module 1 deliverable.

Section 1: where we are now

Answer briefly.

Have we used AI before?

Which tools have we tried?

How often do we use them?

Is AI part of any repeatable workflow right now, or only occasional experimentation?

Section 2: where time and attention are leaking

List 3 to 5 recurring tasks that slow the business down.

Task	How often it happens	Who handles it	Time spent	What makes it frustrating?

Section 3: what AI might help solve

Use the problem-first sentence builder.

I want AI to help me…

So I can…

The inputs/context would be…

The review step would be…

Section 4: what we need to protect

Check any that apply.

Add your own:

Section 5: what a useful result would look like

Be specific.

Desired result	Current baseline	Better result
Example: newsletter drafting time	3 hours per issue	60 to 90 minutes per issue

Completion check

Before moving to Module 2, make sure you can answer:

Completion check What have we already tried with AI? Which recurring tasks are slowing us down? What outcomes would make AI worth the effort? What information or decisions need protection? What would prove that AI is actually helping?

Component 7

AI expectation reset

AI can help you…	AI cannot do this by itself…
Draft faster	Understand your business context
Summarize information	Know what is strategically right
Find patterns in provided data	Fix messy data without guidance
Create first versions of documents	Guarantee accuracy
Repurpose content	Replace your brand judgment
Support repeatable workflows	Repair a broken process automatically
Reduce manual effort	Remove the need for policy and review