A 2-Minute Framework for Finding Your Best AI Opportunity (Without Hiring Anyone)

Before you hire a consultant, read a whitepaper, or sit through a vendor demo, there's a faster way to find your best AI opportunity. It takes about two minutes and requires nothing except a basic understanding of what your team actually does.

The four-question filter

List every task someone on your team does. Then apply this filter:

1. Done weekly or more often
2. Follows the same steps every time
3. Involves copying data between systems or documents
4. Makes people sigh when it's their turn to do it

Rank the tasks that clear all four criteria by hours per week. The top of that list is almost always your first automation.

Why these four criteria and not others

Each criterion is doing specific work.

Frequency matters because automation ROI compounds. A task that takes two hours a week returns 100 hours a year. A task that takes two hours a month returns 24 hours. Same effort to automate, very different payback.

Consistency matters because AI and automation tools require defined inputs and defined outputs. A workflow that's different every time it runs is an engineering project, not an automation candidate. Start with the boring, predictable ones.

Data movement matters because it's the most automatable pattern there is. Export a file, clean it, paste it somewhere, format it, send it — every step is mechanical. None of it requires human judgement. All of it burns time.

The sigh test matters because it's a reliable proxy for cognitive load without value. If a smart person groans when a task lands in their queue, it's because they know their brain isn't actually needed. The task is using their hands, not their head.

What wins this exercise

Reporting and data re-entry win roughly 80% of the time. Across industries, across company sizes. The CFO's Monday morning report that three people touch over Friday afternoon. The weekly operations summary that someone rebuilds in Excel from four different exports. The client-facing dashboard that's actually a manually formatted spreadsheet.

This isn't because reporting is uniquely suited to AI. It's because reporting is where the four criteria all converge most reliably: weekly, structured, involves copying data, and universally dreaded by the person who drew the short straw.

What to do with the list

Once you have the top three candidates, the next question is: which one has clean enough data to actually automate right now? The two-minute exercise tells you where the opportunity is. A half-hour conversation with your team tells you whether the data is accessible or buried in a system with a painful API.

The practical path is:

1. Pick the top task that has accessible data
2. Map the current workflow step by step (15 minutes with the person who owns it)
3. Identify the single step that consumes the most time and has the most defined inputs/outputs
4. Automate that one step before you automate the whole workflow

Partial automation that works is more valuable than full automation that doesn't ship for six months.

If you'd like a structured version of this exercise with scoring across your full operation, the AI Readiness Scorecard.

walks through data maturity, workflow readiness, and where your highest-value opportunities sit.