Use these seven questions to prepare for an AI implementation conversation.
1. What are we trying to improve?
Name the business problem in one sentence. "Help our team prepare consistent project briefs" is more useful than "use AI across the company".
Describe the people affected and what they need. Define an observable result rather than a general promise of productivity.
2. How does the work happen today?
Walk through one real example with the person doing the work. Record the inputs, tools, approvals and exceptions. Pay attention to the informal steps: information copied from an email, a decision remembered from a call, or a spreadsheet only one person understands.
You cannot assess an improvement against a process you have not described.
3. What information will the system use?
List the information required and where it comes from. Confirm who owns it, who may access it and whether you have permission to use it in the proposed system.
If material contains customer, employee or commercially sensitive information, review how the proposed provider handles it before testing. Use invented or appropriately sanitised examples while the requirements are being established.
4. Who is responsible for the output?
Assign an owner. Decide which outputs require review and what the reviewer needs to check.
An AI-generated recommendation should not become a business decision simply because it appears in a dashboard. Make approval explicit where it matters.
5. How will we test whether it helps?
Agree a baseline and acceptance criteria. Select representative examples, including awkward cases and missing inputs.
Measure time spent checking and correcting, output quality and whether the process is usable. Set a review point and define what would cause you to stop the pilot.
6. Can our team run it?
Ask who will maintain access, investigate failures, update instructions and answer questions. Include the people doing the work in the design and test the handover with them.
If the workflow only makes sense to its builder, the handover needs more work.
7. What does it cost beyond the subscription?
Consider implementation, integrations, usage charges, training, review effort and maintenance. Check how pricing changes with volume and what happens if you stop using the tool.
You do not need perfect forecasts to begin. You do need enough understanding to avoid committing to an unmanageable process.
Turn the answers into a short brief
Write one page with five headings: the problem, the current process, the desired result, the constraints and the pilot. Attach one example of a good output.
That brief gives a consultant something concrete to assess. It also helps your team agree what they are asking for before discussing software.
Want to work through it together?
Book a deep consultation with Juan at Triple Pepper. Bring your questions and an example of the work you want to improve. We can explore the opportunities, constraints and appropriate next steps. Scope and fee are agreed before booking.
---