An identifiable result
Once the work is complete, it is clear what has been delivered, and there is a specific person who will use it.
A workflow is usually suitable for a first pilot when it meets all four of these conditions:
Once the work is complete, it is clear what has been delivered, and there is a specific person who will use it.
Someone can decide whether the result is acceptable and can coordinate a response or stop the work when a problem arises.
It is possible to distinguish whether the preceding step's deliverable meets the standard from whether the next step is ready to start now.
After the pilot, another person can be asked to repeat the work using the instructions left behind, with the result assessed against the same standards.
The more fully these four conditions are met, the easier it is to judge whether the improvement was effective after the pilot.
After sales confirms the scope, the delivery team receives the materials. First, check whether the content is complete and meets the agreed requirements. Then separately confirm whether the people and authorization needed to start are in place.
AI can help check required items and formatting; people remain responsible for final acceptance and judgments about exceptions.
Why this is suitable for a first pilot: The result is clear, the downstream recipient is known, and the handoff can be assessed separately for quality and readiness conditions.
Record whether each request reaches the correct handling team on its first routing, who takes over when there is uncertainty, and how that uncertainty is flagged.
One possible division: AI assists with classification and flags uncertain requests; people confirm routing and take over exceptions.
Why this is suitable for a first pilot: Responsibility boundaries are relatively clear, there is a fallback when an exception occurs, and it is straightforward to test whether someone else can repeat the work by following the rules.
Check whether all required fields are complete, route exceptions to a designated person for review, and record whether the entered data is used by the subsequent process.
One possible division: AI assists with extracting fields and flags missing information; people review exceptions and confirm the entered data.
Why this is suitable for a first pilot: The process has clear boundaries, and acceptance and downstream use can be observed separately.
Clarify who will use the materials and what makes them ready for use. After the trial run, ask another colleague to prepare them again using only the instructions left behind.
One possible division: AI assists with organizing and updating drafts; people confirm the content and usage standards, and another colleague tests repeatability.
Why this is suitable for a first pilot: The scope is manageable, and the test directly shows whether another person can reuse the method.
If most of these questions can be answered clearly, this workflow is a reasonable choice for a first pilot.
Compare the pilot with the original process, checking quality, waiting, rework, human intervention and cost before deciding what comes next.
If the evidence supports a change, update roles, manuals, SOPs and AI Skills together, then ask another colleague to try the process.
Bring a recurring task, a recent sample and the result needed downstream. We can begin with this specific issue.
Make sure one workflow can be reliably taken on and tested before considering expansion. To clarify your own business workflow, start with the contact page.
These four illustrative scenarios use realistic workflows to show the full process, from agreeing on scope to testing whether someone else can repeat the work.
What is delivered, and who uses it?
The deliverable is a proposal a client can use to make a decision. The client-facing colleague and the internal delivery team use it to understand the scope, the risks, and whether the next step can begin.
What is agreed before starting?
The pilot covers standard proposals for one kind of client. The proposal writer owns it. Before starting, the team reviews the last 10 proposals to see how many revisions they usually needed after first delivery and how many passed on the first attempt. If the new writer still needs explanations from the original writer after three attempts, the team pauses and holds off on expansion.
What gets checked at handoff?
Check two things separately.
First, does the proposal meet the standard? Is the scope included, and are the risks clear?
Second, can the client-facing colleague use it now? Do they have the materials they need, and has the next step been cleared?
The two checks show whether the proposal needs work or whether the recipient is still waiting for something.
What evidence did the pilot leave?
The work followed the agreed process. More proposals were usable on first delivery. The client-facing colleague used them, and fewer revisions were needed on average.
What happened when someone else repeated it?
Another colleague followed the instructions alone and still needed the original writer to explain how to assess risk. The team added explicit risk criteria. On the next attempt, the colleague completed the proposal independently, so the team kept the method and applied it to similar proposals.
What is delivered, and who uses it?
The deliverable is a ticket with a category and priority so the right person can act on it. The support lead and frontline staff use it.
What is agreed before starting?
The support lead owns the pilot, which covers three common, low-risk ticket types. Before starting, the team records average routing time and the misrouting rate. The pilot pauses if the misrouting rate stays above the baseline for two consecutive weeks.
What gets checked at handoff?
First, check whether the category and priority follow the rules.
Then check whether the assigned person is available and has permission to handle the ticket.
The separate checks show that many delays come from staff availability or missing permissions.
What evidence did the pilot leave?
The team followed the routing rules. The misrouting rate fell. Assigned staff received the tickets and began work, and average time to first response went down.
What happened when someone else repeated it?
A new colleague followed the instructions and generally classified tickets correctly but sometimes skipped the permissions check. The team added that check to the instructions. The colleague then completed the process independently. The team kept the method and continued observing before deciding whether to cover more ticket types.
What is delivered, and who uses it?
The deliverable is an expense variance explanation that a business lead can use in a discussion. The business lead and the finance reviewer both use it.
What is agreed before starting?
The pilot covers monthly variances for one business line, with the financial analyst as owner. Before starting, the team records how complete the explanations are and how often the business lead asks follow-up questions. The pilot pauses if the number of follow-up questions does not fall.
What gets checked at handoff?
First, check whether the explanation includes the amounts, the comparison basis and the reasons that have been verified.
Then check whether the business lead received it and scheduled time to discuss it.
The checks separate missing information from a delay in finding time to review it.
What evidence did the pilot leave?
The agreed process was followed. Explanations became more complete. The business lead used them in meetings, and the follow-up actions were clearer.
What happened when someone else repeated it?
When another colleague repeated the work, some comparison rules proved inconsistent. The team documented a single set of comparison rules, and the colleague completed the next attempt independently. The team kept the method.
What is delivered, and who uses it?
The deliverable is a set of instructions that lets a new colleague carry out a task on their own. The training lead and new team members use it.
What is agreed before starting?
The process owner runs the pilot on one frequently performed task. Before starting, the team records how many new colleagues can finish independently and how often they need to ask questions. The pilot pauses if the independent completion rate does not improve.
What gets checked at handoff?
First, check whether the instructions cover all essential steps and common exceptions.
Then check whether the new colleague has system access and usable data for practice.
The two checks show whether to improve the instructions or provide the missing access or data.
What evidence did the pilot leave?
The instructions followed the process and covered all essential steps. New colleagues used them, and more could finish the task independently.
What happened when someone else repeated it?
Two new colleagues used only the instructions. One still had to ask for help with an exception. The team added criteria for handling exceptions, and the next attempt could be completed independently. The team kept the method and considered applying it to similar tasks.