AI Lab

Turns a variance into a list of causes you can test.

You are an experienced workforce management analyst. A contact-centre forecast for [FICTIONAL queue, e.g. "Billing chat"] missed by [X%] on [day]. Context I can share: [one or two safe facts, e.g. "a public holiday the day before"].

1. List 6 plausible causes, most likely first.
2. For each, tell me which data would confirm or rule it out.
3. Suggest the 3 cheapest checks to run first.
Do not invent numbers. Say clearly what you cannot know.

Check: treat the causes as hypotheses. Verify against your own data before you act.

A tight 30-minute agenda that ends in decisions, not discussion.

Design a 30-minute weekly forecast review for a contact-centre planning team of [number] people. Inputs available: forecast vs actual by day, a list of known events, and last week's actions.
Give me: a timed agenda, the 5 questions the chair should ask, and a one-page template for recording decisions and owners.
Keep it practical and avoid jargon.

Check: trim anything that does not fit your real meeting rhythm.

Generates options and trade-offs so you can choose, not just accept.

I plan schedules for a team of about [number] agents working [shift length] shifts with [one lunch and two short breaks]. Peak demand is mid-morning and early afternoon.
Suggest 3 different ways to place breaks and lunches. For each, explain the effect on coverage at peak, fairness to agents, and how hard it is to run day to day.
Finish with the questions I should ask before choosing one.

Check: confirm local labour rules and any agreements with staff before you change a real break pattern.

Finds the ways a pattern fails before the first Monday it runs.

Here is a described shift pattern (no real names): [e.g. "4 days on, 3 off, starts at 07:00, 09:00 and 11:00, weekend rotation every third week"].
List the 8 most likely ways it breaks in practice, such as absence, public holidays, training weeks and demand spikes. For each, give an early warning sign and a simple fix.

Check: rank the failure modes yourself. You know which ones your operation actually sees.

Practise the structure on made-up data, then reuse the structure on your own.

Below is a FICTIONAL weekly summary table. Write an executive summary in plain English: 5 bullets maximum, then one risk and one decision I need from leadership. Use only the numbers shown.

[paste a fictional table with columns such as week, forecast, actual, service level, shrinkage]

Check: recompute every number it quotes from your table.

One message, three versions: operations, leadership and frontline.

Rewrite this message three ways: for operations managers (detail and actions), for senior leadership (outcome, risk, ask, under 80 words), and for frontline agents (what changes for them and why, friendly and clear).

Message: [your fictional or already-public message]

Check: make sure the versions do not contradict each other, and that nothing was added that is not true.

Helps you structure a coaching chat. Describe the situation in general terms only.

I need to talk with a team member about a pattern in [general topic, e.g. "schedule adherence"]. Do NOT use real names or details.
Give me: an opening line that is calm and specific, 4 open questions, 3 possible root causes to explore before assuming anything, and a way to agree one next step.

Check: never paste real performance data about a named person. Follow your HR process for anything formal.

Imagines the change failed, then works backwards to what to do now.

We plan to [describe a change in general terms, e.g. "move to a new intraday reporting rhythm"]. Imagine it failed badly three months from now.
List the 8 most likely reasons, grouped by people, process and tools. For each, give one thing we can do now to prevent it and one signal to watch for.

Check: pick the three that feel most real to you and ignore the rest.

Understand what you inherited before you change it.

Explain this Excel formula step by step in plain English, say what each part does, and point out anything that could break it (blanks, text in number cells, ranges that do not grow).

[paste the formula only, with generic cell references and no real data]

Check: test the explanation on a few made-up rows before you rely on it.

Gets the structure right before you build.

Design a spreadsheet layout to track shrinkage by week for a team of about [number]. List the tabs, the columns on each, the formulas I need in words, and the 3 checks that would catch a data-entry mistake. Do not write macros.

Check: build a small version with fictional rows first.

Describe the columns instead of pasting the data.

I have an export with these columns: [list column names and types, e.g. "Date (text like 05/11/26), Queue (text), Calls offered (number with commas), Handle time (hh:mm:ss as text)"]. Problems: [e.g. "dates are mixed formats, some numbers are text"].
Write step-by-step Excel (or Power Query) instructions to clean it, and a short list of checks to confirm nothing was lost.

Check: run the steps on a copy and compare row counts and totals.

For team leaders and managers heading into a planning conversation.

I lead a contact-centre team of about [number] people. I meet our workforce management team weekly.
Write 8 questions I should ask about next week's forecast, schedule and risks, in plain language. For each, say what a good answer sounds like and what a worrying answer sounds like. Finish with 3 things WFM will probably want from me.

Check: keep the questions that fit your operation and drop the rest. Ask your own WFM team what they would add.

Helps you reason about class timing before you talk to WFM.

I plan training classes for a contact centre. A class of [number] new hires takes [number] weeks of classroom training and [number] weeks of supported nesting before they are fully productive. Typical drop-out during training is about [percent].
Explain how to work out when the next class should start so that the team is covered when expected leavers go, what information I need from the workforce management team, and 5 risks of starting too late or too early.
Use general terms only. Do not invent numbers.

Check: use your real ramp and drop-out rates, kept on your own systems, when you do the actual planning with WFM.

Fits coaching and calibration into the plan as scheduled activity.

I run quality for a team of about [number] agents. I want each agent to get [number] minutes of coaching a week plus a monthly calibration session.
Suggest 3 ways to place this into the weekly schedule without leaving gaps at peak times. For each, explain the effect on service, the effect on agents, and what I need to agree with the workforce management team first.

Check: agree the final pattern with WFM, because they know where the real peaks are.

Turns a headcount gap into a story that HR, finance and operations can all follow.

Explain in plain English, for a non-specialist audience, why a contact centre needs to hire more people than the size of its headcount gap. Cover attrition, no-shows, training drop-out, the time before new hires are fully productive, and shrinkage.
Use a small fictional example with round numbers, show the working in a short table, and end with 3 questions that the recruitment team should ask the workforce management team.

Check: rebuild the table with your own rates before you present it. The example is only for understanding.

Make the safe habit repeatable, without pasting the sensitive text anywhere.

Create a redaction checklist I can use before sharing a workforce management report or screenshot outside my organisation. Cover names, IDs, client and site details, contract terms, screenshots of internal tools, file metadata and chart axes that reveal real volumes.
For each item, give a safe replacement (for example "use Queue A", "scale volumes by a random factor"). Keep it to one page.

Check: when in doubt, share less. The only safe confidential detail is the one you left out.