AI Engineering
/
AI for Team and Tech Leads
/

Turn an executive's AI mandate into a scoped pilot

AI for Team and Tech Leads · 90 minutes

Turn an executive's AI mandate into a scoped pilot

Answer a COO who wants AI on every customer email and support costs down by a third: find the goal behind the request, check both targets against the team's figures and rules, and commit to a pilot you can measure.

Easy
Conceptual

Scenario

1.5 h

Used in interview rounds:

Leadership Scenarios (Easy)

Step 1: The question

Using the support scenario, explain the goal behind the COO's request and check both targets against the team's figures and rules, showing your arithmetic. Propose a first pilot: which email types, the automation level and human checkpoint, success and stop measures, a comparison, and a decision date. Then write the short reply you would send the COO, saying what you will and will not commit to and why, and say what you would tell the support team.

Scenario: an AI mandate for a support team

The company, team, figures, rules, and prices are fictional.

You lead customer support at a regional internet service provider that sells home broadband. Your team has 45 agents who answer phone, chat, and email, plus 3 team leads and 2 QA specialists. This morning the chief operating officer (COO) wrote to you:

The board wants to hear our AI plan. By the end of next quarter I want AI answering every customer email, and support costs down by a third. Can you own this?

Next quarter ends in 20 weeks.

Team facts (a typical month)

  • Agents log 6,000 handling hours: phone 3,250, chat 820, and email 1,930. The help desk logs handling time (from opening an item to sending the reply) for every email.
  • The support budget is $307,500 a month. Agent time is $246,000 of it; the rest is the phone system, help desk software, training, and management.
  • Email volume grew 18% over the past year with the same headcount. The median wait for a first email reply is 26 hours.
  • Billing emails go to a pod of 10 agents, and the help desk hands them out in turn (round-robin).
  • QA reviews 100 sent email replies a week. About 3 in 100 contain an error: a wrong amount, a wrong date, or a wrong policy.

Email mix (a typical month)

Re-contact is the share of emails where the same customer writes again about the same issue within 7 days.

Email typeEmailsAverage handling minutesRe-contactWhat a correct reply needs
Billing question, no credit asked3,600510%Bill lines and payment dates from the account
Credit or refund request1,200916%A decision on a credit by staff allowed to grant it
Outage report2,400425%Live outage status for the customer's area
Moving home1,2001215%A service check at the new address and an install date
Cancellation900108%Account changes, a final bill, equipment return
Troubleshooting3,0001430%Line test results and router details; often a technician visit
Complaint6002022%The customer's case history

Rules and tools

  • Only staff may grant credits or refunds, each role up to a set limit. No automated system may grant or promise one (company rule).
  • A person must acknowledge every complaint within 2 business days (a company rule in this scenario).
  • Every cancellation goes to the retention group, which decides whether to make a retention offer; no one else may make one.
  • The help desk has macros (fixed reply templates an agent inserts), and a network status page lists known outages by area.
  • Option A is the help desk vendor's AI drafting add-on, at $0.047 per drafted reply (fictional price). It can read the customer's account summary (plan, bill lines, payments) and the knowledge base. It cannot read the status page, line tests, the service check for a new address, or past case history.
  • The vendor's material says agents using its drafts handle emails in half the time. Nobody has measured that on your emails.
  • IT's standard review of a new tool's data terms takes 2 weeks.

Constraints

  • Use only the scenario's figures and rules, and label any assumption as yours
  • Keep the reply to the COO under about 200 words
  • The complaint and credit rules are company rules in this scenario; give no legal advice

Deliverables

  • The goal behind the request and what you would confirm with the COO
  • A check of both targets against hours, budget, and the scenario's rules, with the arithmetic
  • The pilot: email types, automation level and checkpoint, comparison, success and stop measures, decision date
  • A short reply to the COO and the message to the team
  • What you commit to, by when, and what you do not, with a reason for each

Step 2: Hints

Optional. Each hint gives a little more away.

Step 3: Your answer

Kept in this browser. Saving your score below keeps a copy with it.

0 / 5000

Step 4: Reference answer

Step 5: Score yourself

Score each dimension 0, 1, or 2, then save the score to your history. Readiness thresholds are editorial estimates.

Grades your written answer against this rubric with your own AI provider key from Settings. It is a second opinion: the score you save is still yours.

Goal behind the request

What earns a 2: Separates the outcome from the method and number in the request, names what to confirm with the executive, and ties the pilot's main measure to that outcome.

Realistic limits

What earns a 2: Converts volume and minutes into hours and a budget share, tests the target's best case, and names the rules, authority, data access, review time, and unproven claims that limit it.

Pilot scope and measures

What earns a 2: Chooses email types with a reason for each, sets the automation level and human checkpoint, and fixes a comparison, success measures, stop rules, and a decision date before the start.

Communication up and down

What earns a 2: Writes a short executive reply in plain figures and a team message covering what changes, what stays, how to give feedback, and what is not yet decided.

Commitment decision

What earns a 2: States what is committed, by when, and how it will be judged, what is not committed and why, and what result would change the plan.

0 of 10 · Scored 0 of 5 dimensions

Ready at 8 of 10, with no 0 in Realistic limits or Commitment decision

Kept in this browser until you sign up, then it moves to your account.

Complete this level when you can answer an AI mandate with a goal you have confirmed, a target checked against your own figures and rules, a pilot whose measures and decision date are fixed in advance, and a commitment you could keep.

Your saved scores

Practice saves and mock interview answers on this question, newest first.

No saved scores yet. Score every dimension, then save.

Step 6: Follow-up questions

An interviewer may ask these next.

What is the goal behind an AI mandate, and why find it before you scope anything?

Why is time saved not the same as money saved?

How should a lead treat a vendor's claim that its tool halves handling time?

Why set a pilot's stop rules and decision date before it starts?

What should a team hear when an AI pilot starts?

Step 7: Mark complete