AI PATH PRO / TEXT EDITION
Read it and put it to work.
AI Path Pro · 35 minutes + your trial
PROCESS FIRST
Leave with a process map checked by the people doing the work, one justified change and a measurement plan. Complete the workbook with a colleague who performs the task.
Observe. Simplify. Measure.
01 / THE METHOD
See the whole journey of a task.
Choose a recurring task with a clear beginning and a clear definition of done. Take three recent cases and walk through them with someone who did the work. These cases provide an initial view; extend the sample to reflect the variation you need to understand before deciding.
Define the useful outcome.
Name the trigger, the person waiting for the result and what makes the task complete. For example: an incoming request becomes a checked quotation draft. Keep the scope on preparation if that is the work you want to improve.
Follow the work as it happens.
For each step, record the input, output, owner, working time and waiting time. Include copied data, clarification requests and workarounds. Correct the map whenever a real case reveals a different sequence or an undocumented handoff.
Separate the symptom from the cause.
“Quotations take too long” is a symptom. Repeated data entry or missing specifications may be causes. Mark which explanation has evidence from cases or measurements, and which is still an assumption to investigate.
Reduce the work first.
Consider removing a step, simplifying it or agreeing a common working rule. Then consider integration, rule-based automation and AI for any remaining interpretation task. Choose technology to support the change you have identified.
Test one bounded change.
Choose a small change that can be reversed. Compare the same type of case before and after. Include preparation, review and rework in working time; measure waiting separately. Set a quality requirement and a review point.
02 / WORKED EXAMPLE
Fewer loops before the quotation.
A fictional workshop example of routine quotation preparation. Timings are teaching assumptions, and the trial design is a proposed future state. The table does not report customer results or promise savings.
| Part of the work | Current assumption | Proposed design |
|---|---|---|
| Data entry | 12 minutes: the same data in two places. | 0 additional minutes: one shared record. |
| Clarifying missing details | 17 minutes: an unstructured request. | 7 minutes: an intake form with required fields. |
| Arranging the handoff | 10 minutes: manual status chasing. | 3 minutes: visible status and ownership. |
| Preparing and checking content | 14 minutes: completing the substantive work. | 14 minutes: the quality check stays. |
A difference of 29 minutes per case.
The assumptions give 53 − 24 = 29 minutes. This is a calculated difference between two hypothetical ways of working. Actual working time must be measured during the trial.
Waiting is a separate measure.
If waiting falls from 24 hours to 2 hours in the example, the difference is 22 hours. Do not add that to the 29 minutes of working capacity: they measure different things.
Capacity needs a practical purpose.
For 50 comparable cases a month, the assumed difference is 50 × 29 ÷ 60 = 24.2 working hours. This is not yet a financial saving. Decide how that capacity would be used.
03 / YOUR PROCESS
One page. One workable change.
The examples show possible answers. On your own sheet, name real cases, owners and verifiable data. When something is unknown, state how you will find it out and who can help.
- 01 / START AND FINISH
What triggers the work, and exactly which output marks its end?
Example answer: Incoming request → technically checked quotation draft. Sending the quotation is a separate process.
- 02 / RECIPIENT AND DONE
Who needs the result? Which fields and quality conditions are mandatory?
Example answer: The salesperson needs an itemised draft with sources; missing specifications remain visible.
- 03 / CURRENT WORK
Which steps, handoffs and waits occur? What did you measure, and what did you estimate?
Example answer: Time logs for five routine cases; rework included. Waiting recorded in a separate column.
- 04 / CAUSE AND EVIDENCE
Which specific obstacle would you change? What evidence shows that it matters?
Example answer: Three of five requests lacked dimensions; each needed another clarification round.
- 05 / SMALLEST CHANGE
What will you remove or simplify? Which control must remain?
Example answer: A common intake form with a required dimensions field. The trial retains the technical review.
- 06 / TRIAL AND DECISION
Who leads the trial, what will they measure, and when and how will they decide?
Example answer: Owner: designated sales manager. Next ten routine cases; working time, errors and clarifications; a joint review.
04 / PUT IT TO WORK
Your next 35 minutes.
Turn the workbook into a shared decision. Use these three checks, choose one trial and agree when you will review its result.
- The map reflects real work
Ready: Someone who performs the task has checked the steps, including workarounds.
Still to do: You only have the official flowchart. Walk through an actual case.
- The cause has evidence
Ready: The selected obstacle is supported by a specific case, measurement or observation.
Still to do: The team only suspects the cause. Gather targeted observations to test the assumption.
- The trial can be evaluated
Ready: The owner, comparison, quality requirement and rollback method are clear.
Still to do: There is only a speed target. Add a quality measure and a review point.
- 10 MINUTES
Put three cases on the table.
Select completed tasks of the same type. Examine the original input, final output and intermediate exchanges. Mark missing measurement data rather than filling gaps with assumptions.
- 15 MINUTES
Map it with the person doing it.
Draw the work in its actual sequence. Circle repetitions and waits. Choose one obstacle whose cause you already understand, and check that the proposed change addresses that cause.
- 10 MINUTES
Agree one small trial.
Complete the final two worksheet fields. Ask the process owner to name the participant, the comparable cases, the measurement and the review point. Keep the first trial small enough to learn from.
Related sources
- Make: Scenarios
Use this when the simplified task can be automated with stable rules.
- v0 documentation
For designing a dedicated intake form or working interface around the simplified process.
A better process can remove the need for the automation you planned.