Before you automate, ask the people who fix the mistakes.
Use frontline knowledge to define the problem and compare practical options.
Listen before choosing the solution
A fictional support team wants AI to answer refund questions faster. The employees handling those questions point to conflicting instructions as a source of repeat contacts.
That observation changes the proposal. The team can compare clearer rules, AI assistance and a combination of both before deciding what to fund. Frontline knowledge helps define the work the pilot should improve.
Build a useful pilot
Clarify the rules first. If an assistant is still worth testing, keep the pilot bounded, preserve human review and name the person responsible for exceptions.
Record a baseline for repeat contacts, resolution time and answer quality. Use the same definitions during the pilot and include checking and rework. No improvement has been measured in this illustration, so there are no savings or performance results to report.
Connect evidence to the decision
Compare options using their full costs and observed results. Distinguish useful capacity from a demonstrated reduction in expenses. Decide whether to expand, redesign or stop the proposal.
The AI Investment and Business Case Toolkit helps teams connect assumptions, pilot evidence and the next funding decision. Start with the people who understand where the process breaks down.
Read the video transcript
Before you automate the process, ask the people who fix its mistakes. In this fictional customer-service team, leaders want AI to answer refund questions faster.
The people handling the tickets point out a different problem. Customers receive conflicting refund instructions. They contact support again because the answer changes depending on which page they read.
That changes the project question. Is the best first step an AI assistant, clearer refund rules, or both? The team maps the repeat contacts and identifies where the instructions disagree.
They clarify the rules first. Then they can test whether AI helps staff find the right answer. A person checks the recommendation, and unusual cases go to someone who can resolve them.
Before the pilot, record repeat contacts, resolution time and answer quality. During the pilot, measure the same things, including checking and rework. Faster replies matter only if the problem improves.
Compare the options, full costs and results. Then decide whether to expand, redesign or stop. Time released is useful capacity; it becomes a cash saving only when expenses actually fall.
Anvaylo's AI Investment and Business Case Toolkit helps your team connect assumptions, pilot evidence and a funding decision. Start with the people who understand the work.
Download this toolkit, and explore others like it, from the Resources page at anvaylo.com/resources.
Explore the related executive guide and companion working records.
Explore the toolkit →