# AI Opportunity Map

Nick McGuire | nickmcguire.ai

Find one place where AI could make the work easier. Start with the job, then decide what belongs in the solution.

Leave with one worthwhile pilot, a way to judge it, and a reason to stop if it adds more work than it removes.

List three recurring jobs. Compare the friction, the quality of the information, and the cost of a mistake. A frequent task with clear inputs and a checkable output is a stronger starting point than a broad ambition to automate the business.

## Work through it

### 1. Name the work.

Describe a task with a beginning and an end: turn weekly project updates into a report, or extract approved fields from a document. Record who does it, how often, and who uses the result.

### 2. Count the whole job.

Measure a normal run, including gathering inputs, checking, corrections, and handoffs. Estimate monthly effort as runs per month × minutes per run. Label estimates and replace them with observations.

### 3. Check the conditions.

Can you access the inputs legitimately? Are they consistent enough to use? Can a person verify the output? Who owns exceptions? Missing permissions or an unchecked high-consequence decision is a reason to pause.

### 4. Choose a small test.

Prefer meaningful recurring friction, accessible inputs, checkable outputs, and a willing owner. Resolve dependencies first. Test one task on representative examples, including incomplete or contradictory inputs.

## A worked example: A weekly operations report

Fictional worked example. A small services team gathers updates from four project leads. The owner chooses a draft report for review, rather than automatic distribution.

**Current work**
Four runs per month × 90 minutes = 360 minutes per month. This includes collecting updates and correcting the report.

**Pilot boundary**
Use approved updates from one team. Draft completed work, blockers, and next actions. Link each statement to its source. Flag missing updates instead of filling gaps.

**Acceptance check**
The owner checks every factual statement against the source. Missing information is visible. Nothing is emailed until approved. Include a stale update and conflicting dates in the test set.

**Illustrative effort**
If each run takes 20 minutes to prepare and 25 to review, four runs take 180 minutes. Add 60 minutes of monthly maintenance: 240 total. Compared with 360, that leaves 120 minutes of potential monthly capacity, before one-time setup.

**Decision**
Measure actual effort before expanding. Two hours of capacity is not automatically a cash saving. Include subscriptions and setup effort when judging whether the pilot is worthwhile.

## Your worksheet

### Three recurring jobs

For each: task, owner, frequency, user of the output, and where it gets stuck.

________________________________________

### Current effort and evidence

Runs per month × total minutes per run. What was measured, and what is estimated?

________________________________________

### Inputs, access, and consequences

Where does the information live? Who may use it? What happens if the output is wrong?

________________________________________

### The first pilot

Pick one job. Explain why it comes before the other two. Name any dependency to resolve first.

________________________________________

### Acceptance and review

What must the output contain? Who checks it? Include missing data and conflicting input examples.

________________________________________

### A decision you can revisit

Record setup effort, recurring tool cost, preparation, review, and upkeep. Set a review date and a stop condition.

________________________________________

## Decide what happens next

### Start

The job repeats, the input is available, the output can be checked, and someone owns the result. Run a bounded pilot.

### Prepare

The opportunity is useful, but the source, permission, process, or owner is unclear. Fix that dependency first.

### Leave it alone

The task is rare, the cost of an unchecked error is high, or a simpler process change removes the friction.

30 minutes with Nick. Practical advice. No obligation.

https://nickmcguire.ai/resources/ai-opportunity-map
