AI prompts that make frontline managers better leaders, not just faster ones.

Use when: The last ten minutes of your shift.

Here are my rough notes from today's shift: [paste notes with team member names replaced by roles or initials and guest details removed]. Turn them into a 60-second handoff for the next manager: 1) what is still open and who owns it, 2) the top 3 risks for the next shift (staffing, equipment, low inventory, large orders), 3) anything a person must verify in person, and 4) one team member who deserves a thank-you and why. Keep it under 120 words.

Data check: No guest names or contact details. No health reasons for absences. Book: Ch. 5, 6

Use when: Before a busy daypart.

Plan a 5-minute pre-shift huddle for [daypart] given: forecast [busy or normal], focus area [e.g., order accuracy, drive-thru times, catering], and new team members on shift [number]. Give me one opening line, one standard to reinforce with a real example, one question for the team, and one way to recognize someone specifically. Make it sound like me talking, not a memo.

Data check: Green. No personal information needed. Book: Ch. 13, 16

Use when: When people call out and the shift still has to run.

Two people called out for [shift]. Here is who is available, by initials and position, with their hours this week: [list]. Our rules: [minors' hours, maximum hours, breaks, overtime approval]. Give me 2 or 3 coverage options with the tradeoffs: overtime, fatigue, skill coverage, and fairness. Flag anything I must verify with payroll or local labor law. Then remind me who to thank, and who may need a check-in rather than a schedule fix. I make the final call.

Data check: Leave out the reason for a call-out if it is medical or family-related. Book: Ch. 9, 10, 12

Use when: Before a new team member's first shift.

Design the first week for a new [position] starting [date]. Separate it into what can be automated (forms, reminders, training modules, schedule notices), what a trainer does, and what I must do personally, including a one-on-one welcome meeting. For my welcome, give me 3 questions to learn who they are and one sentence explaining what we value here.

Data check: Red: never put work-authorization documents, Social Security numbers, or IDs into a general AI tool. Book: Ch. 4, 10

Use when: When someone has been struggling for more than a shift or two.

A team member in [position, time in role] has been struggling with [observable behavior, e.g., missing ticket times on the grill during lunch] for [how long]. What I have seen directly: [facts]. Help me prepare, not script: separate facts from my assumptions, give me 3 questions to understand what is getting in the way, one clear expectation in plain language, and one specific support I can offer. Then role-play their likely first reaction so I can practice staying calm and clear.

Data check: Describe behavior, not the person. No names or health details. Book: Ch. 13

Use when: After a complaint or a low-star review.

A guest complaint: [paste it with name, phone, email, and order number removed]. Verified facts: [list]. What I can offer within policy: [options]. Still uncertain: [list]. Draft a short, non-defensive response that sounds like a real local leader, does not admit to anything we have not verified, and states the next step. Then give me an internal list: what to fix in the operation, who owns it, and whether any team member needs coaching or deserves recognition. Flag anything involving allergies, food safety, injury, or discrimination for escalation instead of a reply.

Data check: Red: confidential guest data stays out. Remove names and contact details first. Anatomy of an AI Workflow Book: Ch. 6

Use when: When you notice someone doing it right.

Here is something I noticed this week: [behavior, situation, who it helped]. Help me make the recognition specific: the exact behavior, why it mattered to guests or teammates, and how to say it in front of the team in under 30 seconds. Also suggest one follow-up question to ask privately. Do not make it sound like a template, and do not guess at their feelings.

Data check: Green. Initials are enough. Book: Ch. 12, 16

Use when: When a report flags overtime, labor variance, or a late clock-in.

My system flagged this: [e.g., overtime on Tuesday, a labor-cost variance, a late clock-in], with no names. Before I treat it as a cost problem, help me ask what human story might sit behind it. Give me 3 questions to find out what actually happened, who might need support, and who might deserve recognition. Remind me to confirm the facts with people, not just the report.

Data check: Describe the alert, not the person. Never paste payroll records. Book: Ch. 12

Use when: Before you address a pattern of call-outs or tardiness.

I need to talk with a team member about [attendance or tardiness pattern: dates and facts only]. Help me prepare a conversation where the feedback is actually received and understood: an opening that is direct and respectful, the specific facts, the expectation, a question that invites their side, and how I will confirm they understood. Remind me that I make any disciplinary decision and that I follow our documented policy.

Data check: Leave out medical or family reasons. Those stay in person and in the approved HR process. Book: Ch. 11

Use when: Before someone runs their first shift.

I am considering giving [role or initials] their first [opening, closing, or drive-thru lead] shift. Based on what I have observed: [list], what should I verify before giving independent authority? Suggest a supervised practice shift, the boundaries and escalation rules, 3 coaching questions for the debrief, and how I will recognize their growth with specific evidence.

Data check: Observed behavior only. No HR records. Book: Ch. 15

Use when: Right after AI takes a task off your plate.

AI just gave me back [number] hours a week from [task]. Before those hours disappear into more administration, help me decide where they go. Give me 3 specific reinvestments in people (who by role, what kind of time, how often) and one way to check in 30 days whether the time actually reached them.

Data check: Green. Roles only. The Capacity Dividend Book: Ch. 3, 12

Use when: Before your team starts using AI tools on its own.

Help me write a one-page "how we use AI here" guide for my frontline team at [type of business]. Include: approved tools, what never goes into AI (guest information, payment data, other team members' personal details), when to ask a manager first, how to double-check AI answers, and what we will always do in person. Keep it at an 8th-grade reading level and short enough to post in the break room.

Data check: Have the result reviewed against your organization's actual policy. The Frontline AI Data Rule Book: Ch. 10, 17

Use when: Before you change a workflow your team depends on.

I want to introduce AI to my team in a 10-minute meeting without causing fear. Given [team type, current tools, the one workflow we are changing], help me explain what is changing, what is not, what AI will and will not do, how it gives time back to people, and how anyone can raise a concern. Include 3 questions to ask the team and one commitment I will make to them.

Data check: Green. No personal information needed. Book: Ch. 17

Use when: When you are doing work someone else could learn to own.

Management keeps work running; leadership grows people. Here is my week: [list of tasks I personally handle]. What am I doing that a capable team member could own with the right preparation? Help me pick one person (role or initials) and one capability, and give me the stretch responsibility, the boundaries, and the follow-up date to schedule before the conversation ends.

Data check: Roles or initials only. Book: Ch. 2

Use when: When the admin is eating your shifts.

Here are the recurring tasks that fill my shifts and my week, with rough minutes for each: [list, e.g., schedule changes, prep lists, review responses, training checklists, inventory counts]. For each, ask first whether it should exist at all, then whether it can be simplified or standardized before anything is automated. Rank the 5 best candidates for AI help, say what a person must still check, and name one block of time I should protect for leading people instead.

Data check: Describe tasks, not people. Book: Ch. 1, 5

Use when: Before you claim AI is saving you time.

For this AI-assisted task: [task, current minutes per week], estimate the real Capacity Dividend. Start with the claimed time saved, then subtract the time for prompting, reviewing, verifying, correcting, handling exceptions, maintaining the workflow, and recovering from failures. Show your math, label estimates, and tell me the usable minutes left. Then ask me where those minutes will go and who they will reach.

Data check: Green. Describe the task only. The Capacity Dividend Book: Ch. 3

Use when: When a prompt works and you want it to work every time.

Turn this recurring process into a responsible AI workflow: [process, e.g., responding to 1 to 3 star reviews]. Map it using seven elements: Trigger, Inputs, Machine work, Human judgment, Action, Exception, and Measure and improve. For each element, say who owns it, what information the machine actually needs (and what it should never see), and what happens when the output is wrong.

Data check: Yellow: design the workflow with sample or sanitized data first. Anatomy of an AI Workflow Book: Ch. 6

Use when: Before you automate any responsibility.

Break this responsibility into its steps: [responsibility, e.g., onboarding a new team member]. Place each step in one of four categories: Human Only, Human + AI, AI + Human Review, or Safe to Automate. Explain why each boundary exists. Flag any step where human presence is the point (a welcome, an apology, recognition) and keep it human.

Data check: Green. Describe the steps, not the people. The Human + AI Work Matrix Book: Ch. 4

Use when: Before and after you add AI to a process.

Help me build a one-page before-and-after scorecard for this AI-assisted process: [process]. Use the same unit before and after (minutes, errors, cycle time, guest or team impact). Label every number as measured, estimated, or unknown. List the claims I can support, the claims I cannot, and the specific place any usable capacity went.

Data check: Use totals, not individual employee data. The Capacity Dividend Book: Ch. 7

Use when: When you are already sure.

Here is the decision I am considering: [decision and my reasoning]. Argue against my current position. Identify assumptions I may be making, information I may be missing, people who may experience unintended consequences, and what evidence should change my mind.

Data check: Describe affected people by role. Book: Ch. 8

Use when: When an AI recommendation looks right on paper.

Here is an AI recommendation: [paste]. Make three columns: what the system knows, what I know from the floor, and what neither of us knows. Mark each decision-relevant item as verified, unverified, or unknown. Then tell me the one unknown I should reduce by talking to a person or watching the work before I act.

Data check: Green or Yellow depending on the recommendation. Remove names. Book: Ch. 9

Use when: Before a change you cannot easily undo.

Assume this decision produces a bad outcome six months from now: [decision, e.g., a new labor model or a drive-thru process change]. Give me the five most plausible reasons why, the warning signs I should watch for, and what I should investigate before proceeding.

Data check: Green. Describe the decision, not individuals. Book: Ch. 8

Use when: Before you paste anything into an AI tool.

Before I use AI for this task: [the task, the tool, and a description of the information I plan to include], classify each piece of information as Green (public, fictional, sanitized, or approved for this tool), Yellow (internal documents, employee situations, schedules, operating data: pause and check approval), or Red (credentials, payment data, health information, sensitive HR records, confidential guest data: stop and escalate). Tell me what to remove, how to sanitize the rest, and whether this task should stay outside the tool. Do not ask me to paste anything Red.

Data check: Run this on a description of the data, not the data itself. The Frontline AI Data Rule Book: Ch. 10

Use when: Before you act on any AI answer that matters.

Audit this AI output: [paste]. Separate it into facts, assumptions, interpretations, and recommendations. For each fact, tell me how I could verify it. Show me where the answer sounds more confident than its evidence, and what I would need to see before I decide.

Data check: Check the output for names or personal details before pasting it anywhere else. The Judgment Gap Book: Ch. 8

Use when: When AI, or an AI agent, is part of a decision.

For this AI-assisted decision or workflow: [describe it], map the Responsibility Chain: who or what produces the output, who verifies it, who decides, and who owns the consequences afterward. If an AI agent can take actions, define its access, required approvals, reversibility, escalation path, and monitoring. Flag any place where someone has responsibility without authority, or authority without responsibility.

Data check: Green. Use roles, not names. The Responsibility Chain Book: Ch. 11