In this article
You have found a task that AI can help you complete. Perhaps it turns rough notes into a usable checklist or helps you prepare the first draft of a customer guide. The result looks promising. Your next step is to investigate where it might be useful.
A small experiment gives that idea a shape. You choose someone to help, make one example, and invite feedback on a specific task. You can learn from that exchange before spending weeks building a website or adding features.
Start with a person and a situation
Describe the moment when someone might need the output. What are they trying to finish? What information do they already have? What makes the task difficult? A narrow situation gives you something you can investigate in a conversation.
Consider a fictional example: an independent photographer preparing for a client kickoff call. They have answers from an intake form, but the information is scattered. Your proposed service turns those answers into a reviewed preparation sheet with decisions, missing details, and questions for the call.
Make one example you can explain
Use fictional or permitted information for the first version. Keep the original input, your instructions, the AI draft, and your corrections. That record helps you explain what the workflow does and where your judgment contributes.
For the photographer example, you might create a two-page sample with a clear label: fictional demonstration. Include an incomplete answer so you can show how you handle uncertainty. A useful sample makes the proposed service easier to discuss without pretending a client has already bought it.
Ask about the work behind the reaction
When someone agrees to review your example, ask how they prepare today. Which information is hardest to collect? Which part of your sample would they use? What would they change? Listen for the actual job they need completed.
A compliment is encouraging. It leaves several questions open: whether this task happens often, whether the result fits their process, and whether they would pay for help. Record the person's words and the next step they choose instead of treating enthusiasm as a sale.
Keep a simple experiment record
Write down the time you spent preparing, creating, checking, and revising the sample. Record any tool costs and what happened when you shared it. If you discuss a paid pilot, state the deliverable, price, timing, revisions, and responsibilities before accepting the work.
Choose a review date for your experiment. You might decide to try a second example, narrow the audience, change the output, or stop. Each decision should come from something you observed. The goal is to make the next investment of time more informed.
Give yourself a place to begin
The Wealth Experiment's free Two Mondays workshop is a place to shape a focused offer, work on a small first version with AI, and plan a customer test. The November 2 and 9, 2026 sessions run from 7–9 p.m. Eastern. Optional setup help and group office hours are included.
You can request a free place on the workshop page. Ashley reviews requests and confirms availability separately. Software, hosting, and usage costs may be separate. Bring an idea, a skill, or a project you have started, and choose one useful question to investigate.
Request a free Two Mondays place
Improving a workflow in a business you already run? Explore the Agency team working lab. These are different starting points: business experimentation here, team workflow practice at Agency.
Your next move
Choose one potential buyer, create a labeled sample, and record what you learn from a conversation.