Logline: A firm buys an AI assistant to answer customers faster: and discovers that the customers were not waiting for answers so much as waiting for the company to remember them.
The Case
A service leader asks for an AI assistant to answer customer questions faster after a difficult quarter. The team has already chosen software, but customers, frontline staff, and leaders mean different things by faster and helpful.
Cast
| Participant | Assigned Lens |
|---|---|
| Peter Drucker | customer reality and business usefulness |
| Jane Jacobs | lived systems and local knowledge |
| Richard Feynman | plain-language explanation and curiosity |
| Mary Parker Follett | participation, power-with, and constructive conflict |
Stakes
- Customers are repeating information across handoffs.
- Frontline staff fear the tool will standardize away judgment.
- Leadership needs relief quickly and has framed the request as a software decision.
The Story
Cold Open - The Bell
A help-desk counter stands alone beneath a bright service light. On it: a brass bell, three half-completed forms, and a sealed carton stamped SOLUTION.
The bell rings. No one comes. It rings again.
HOST: The first ring belongs to a customer.
A second bell answers from behind the curtain.
HOST: The second belongs to the person trying to help them.
A third bell rings from inside the sealed carton.
Richard Feynman: That is either the future or very determined packaging.
Peter Drucker: Does the box know why the customer rang?
Jane Jacobs: Does the company?
Mary Parker Follett: There is our first disagreement. Everyone wants a faster answer. Nobody has agreed on the question.
Scene One - Twelve Seconds
HOST: A service firm has had a difficult quarter. Response times are up. Customers repeat themselves across handoffs. Employees search three systems before answering ordinary questions.
HOST: Leadership has selected an AI assistant. In the demonstration, it answers in twelve seconds.
Richard Feynman: Answers what?
HOST: Policy questions. Status questions. Common troubleshooting questions.
Richard Feynman: I asked what, not which category.
Peter Drucker: He means: what changes for the customer?
HOST: They receive an answer faster.
Jane Jacobs: The customer did not say, “Your answer arrived slowly.” The customer said, “Every person is polite, but the company keeps forgetting why I called.”
Mary Parker Follett: Who heard that?
HOST: A frontline representative recorded it in a service note.
Mary Parker Follett: Was that representative in the software-selection meeting?
HOST: No.
Mary Parker Follett: Then the company invited the data and excluded the knower.
Scene Two - Faster at What?
Peter Drucker: I will defend the leaders for a moment. They see a measurable problem and a plausible instrument. Waiting has a cost.
Jane Jacobs: I will defend the counter. The measurable problem is what survived the trip upstairs.
Peter Drucker: Response time matters.
Jane Jacobs: Certainly. So does whether the fourth person knows what the first three promised.
Richard Feynman: Let us perform a small miracle and use ordinary words. A customer wants the company to remember the problem and carry it forward. The company proposes a machine that talks quickly.
Peter Drucker: That is unfairly concise.
Richard Feynman: Concision is often accused by details that were hoping to hide.
Mary Parker Follett: There are two kinds of speed in this room. Speed of reply and speed of progress.
Peter Drucker: Good. Now we can measure something worth arguing about.
Jane Jacobs: Not yet. First we walk the handoff.
Scene Three - The Walk
The counter unfolds into three small desks: INTAKE, SPECIALIST, RESOLUTION. Each has its own screen. None can see the others.
HOST: Intake records the problem in the customer system. A specialist investigates in a technical system. Billing records exceptions elsewhere. When ownership changes, each person creates a fresh summary.
Jane Jacobs: There is the neighborhood. Three respectable buildings and no street between them.
Richard Feynman: The assistant will search all three?
HOST: Two at launch. The billing system requires a later integration.
Richard Feynman: Then its twelve-second answer may be missing the expensive part.
Peter Drucker: Or confidently summarizing an incomplete company.
Mary Parker Follett: And who corrects it?
HOST: The service representative.
Mary Parker Follett: The person whose work the business case says will be reduced.
The sealed carton rings again.
Jane Jacobs: It has excellent timing.
Scene Four - The World Enters Sideways
HOST: A current signal complicates the case. AI discussions are moving beyond model capability toward the operating structure around it: governed context, evaluation, cost controls, handoffs, and explicit authority. FCG’s recent Signal Desk put the principle plainly: better AI starts upstream.
Richard Feynman: “Upstream” is useful only if we can point to the river.
Jane Jacobs: We can. The river begins when the customer first explains the problem. Every handoff either carries that context or spills it.
Peter Drucker: So the fashionable language is sticky here. Not because “harness” is a fine new word, but because the assistant’s value depends on work the demo did not show.
Mary Parker Follett: Context is not merely documents fed to a system. It is also who may interpret, correct, and decide.
HOST: The signal does not select the tool. It makes one fact impossible to ignore: the assistant inherits the quality of the work around it.
Scene Five - Open the Box
HOST: The leadership team expects a recommendation: approve or reject.
Peter Drucker: Approve the problem. Withhold approval from the proposed definition of it.
Richard Feynman: That sentence needs an instruction manual.
Peter Drucker: Very well. Do not buy speed before deciding what must move faster.
Jane Jacobs: Spend one week on the hardest real handoff. Put intake, the specialist, billing, and the customer’s experience on one wall.
Mary Parker Follett: Not as witnesses to a design already chosen. As co-authors of the work.
Richard Feynman: Then ask three embarrassingly plain questions. What problem are we solving? What else could solve it? Which small change would tell us whether we are right?
HOST: The carton is opened.
Inside is not a machine. It contains four blank cards: REMEMBER, ROUTE, EXPLAIN, DECIDE.
Jane Jacobs: The assistant may earn one of those jobs. It does not automatically own all four.
Peter Drucker: Remember and explain may be useful. Route requires ownership. Decide requires authority.
Mary Parker Follett: And the customer’s job is not “receive generated text.” It is “regain momentum without starting over.”
Richard Feynman: There. We finally opened the box and found the work.
Scene Six - The One-Week Bet
HOST: The room does not reject AI. It refuses to let AI define the problem merely by being available.
HOST: Monday’s move is a one-week redesign of one difficult customer handoff.
Jane Jacobs: Observe the real path, including the side doors and workarounds.
Mary Parker Follett: Include the frontline representative, the specialist, billing, and a customer-safe representation of the customer’s experience.
Richard Feynman: Define one experiment. Can shared context prevent the customer from repeating the story without creating a new correction burden?
Peter Drucker: Measure progress, not performance theater: fewer restarts, clear ownership, resolution quality, employee correction time, and customer confidence.
HOST: If the assistant helps that bounded job, test it. If a process change solves the larger part, make the process change. If both are needed, stop pretending the purchase was ever the strategy.
The customer bell rings once. This time, all three desks light together.
Mary Parker Follett: Integration is not everyone speaking at once. It is the problem no longer being abandoned between them.
Peter Drucker: I began by asking what result changes. I would revise the question.
Richard Feynman: I have been waiting all evening for this humiliation.
Peter Drucker: What work must change so the result can change honestly?
Jane Jacobs: Better.
HOST: A tool request is often a problem wearing a procurement badge.
Field Note From The Arena
- The case: A service leader asks for an AI assistant to answer customer questions faster after a difficult quarter. The team has already chosen software, but customers, frontline staff, and leaders mean different things by faster and helpful.
- What the room initially misunderstood: It accepted buy the assistant now or fall behind as the governing choice.
- The changed frame: The requested tool is not the work. The real job is helping a customer regain momentum without losing context, dignity, or a clear human owner.
- The practical move: Define the problem in the customer’s language, stretch beyond the requested tool, and choose the smallest intervention that improves real progress.
- Memorable line: A tool request is often a problem wearing a procurement badge.
- Method revealed: Everyday Innovation, supported by Jobs To Be Done.
About the voices
The workplace and dialogue are original composite dramatization. Named thinkers are archetypal lenses; the dialogue is not historical quotation. Proposed tests and outcomes remain part of the story.