Producepak logoPak-Bot

Pak-Bot is your fresh produce AI chatbot.

Pak-Bot prompt library: example questions that work

Copy these prompts, change the customer, product or date, and ask Pak-Bot. Short, specific prompts get the best answers.

Book a Pak-Bot demoGet started
Pak-Bot, the Producepak AI assistant robot, ready to answer a question

How to use this prompt library

Pak-Bot answers questions written in plain language, so there is no special syntax to learn. Still, some prompts work better than others, and it helps to see examples. This library collects more than 100 example prompts for fresh produce businesses, grouped by topic. The first ten come straight from the Pak-Bot guide supplied with Producepak; the rest follow the same patterns.

To use a prompt, copy it into the Pak-Bot chat box and change the customer, product, site, grower or period to match your business. Names should match how records are set up in your Producepak account. Pak-Bot only returns data your role and site allow you to see, so some prompts (sales prompts, for example) will only work for users with the right access.

The ten starter prompts

These ten prompts appear in the Pak-Bot guide. They are a good first test of Pak-Bot on your own data.

Two more prompts are shown in Pak-Bot screenshots: chart total sales to top five customers this year and List the 10 oldest inventory items.

Screenshot of the Pak-Bot chat window with the prompt 'List the 10 oldest inventory items' and a results table of ten inventory rows showing part name, units, quantity and manufacture date, with star rating and comment controls below and a message box with microphone button
A Pak-Bot answer with its rating stars, comment box and the microphone button for spoken prompts. Example data.

Anatomy of a good prompt

The most reliable prompts contain some or all of these parts:

PartExamples
Actionchart, list, total, compare, email, print
Measuresales, units, weight, value, yield, rejection rate, orders
Subjecta product, customer, grower, supplier, site or packaging type
Breakdownby customer, by product, by state, by site, by week
Periodtoday, last week, last month, Q1 2026, this year

"Chart sales of avocados by customer last month" contains all five. Short prompts work too ("Total sales last month"), because Pak-Bot fills sensible defaults.

Inventory prompts

More detail: AI inventory answers.

Sales prompts

More detail: sales analytics.

Order prompts

Pak-Bot can also convert customer purchase order documents into sales orders. See document to order.

Quality control prompts

Yield prompts

More detail: quality and yield.

Example output for the prompt chart yield by grower last 30 days: a bar chart ranking five growers from 91 percent down to 64 percent packout yield, using illustrative data
Example output for a yield prompt. Illustrative data.

Supplier, grower and customer prompts

Employee prompts

Pak-Bot knows which employees work at which site and which employees can access sales data. Employee prompts follow the same permission rules.

Email and print prompts

Label prompts

Once your customers' label designs are loaded into Producepak, Pak-Bot can generate those labels instantly, with no need to use the label designer.

More detail: labels, email and print.

Chart prompts

Pak-Bot answer to 'chart total sales to top five customers this year' showing a pie chart and table of Walmart, Loblaws, Woolworths, Tesco and Costa Rica Fruit Co sales totals
Output of a chart prompt. Example data.

Prompt workflows: chains of questions

The best use of Pak-Bot is often a short chain of questions, each building on the last.

Morning stock check (warehouse)

  1. Total of inventory at site Closters
  2. List the 10 oldest inventory items
  3. Print the 10 oldest inventory items

Customer review preparation (sales)

  1. Walmart orders vs sales Q1 2026
  2. Which Walmart orders were short-shipped in Q1 2026?
  3. Email Walmart orders vs sales Q1 2026 to me

Supplier investigation (quality)

  1. Rejection rate by supplier this quarter
  2. Most common defects for the worst supplier this quarter
  3. Chart that supplier's rejection rate by week

Adapting prompts to your business

The prompts on this page use example names: Angela Coles, ACE Farming Group, site Closters, Woolworths, Walmart, Loblaws, Tesco and Costa Rica Fruit Co. Replace them with your own people, growers, sites and customers. Products should also match your catalogue. If you pack berries rather than potatoes, "Total yellow potato in stock" becomes "Total strawberries in stock" or "Total blueberries in 125 g punnets in stock". The structure of the prompt stays the same; only the names change.

Periods can be adapted too. Replace "this year" with "this season", "last month" with "last week", or "Q1 2026" with an exact date range such as "1 March to 15 April". Explicit dates are the safest choice when a period could be read in more than one way, for example when your financial year does not start in January.

Building your own prompt library

Start a shared document with three columns: role, prompt and what the answer is used for. Add the prompts from this page that apply to your business, rename them with your own customers, products and sites, and test each one. Remove any that do not suit your operation. Add new prompts as people discover them. After a season, the list becomes a practical reference of how your business uses data, and a quick way for new staff to learn which questions matter.

A good library is short enough to read in a few minutes. Ten to fifteen prompts per role is usually enough. Mark the prompts people use daily so new users know where to start, and review the list at the start of each season when products, customers and programs change.

Prompts for the end of the season

When a prompt does not work

Prompt FAQ

Do I need to type prompts exactly as written here?

No. Pak-Bot understands natural language. These examples show patterns that work well; change the wording and names to suit your business.

Can I speak prompts instead of typing them?

Yes. The Pak-Bot chat box has a microphone button for spoken questions.

Why did a sales prompt return no figures?

Pak-Bot only returns sales data to employees whose Producepak role includes sales access.

Prompts by role

The categories above are organised by subject. Many people find it easier to start from their own role. Here are suggested starting sets.

Owner or general manager

Sales manager

Warehouse and dispatch

Quality and food safety

Grower relations and purchasing

Words that change the answer

Small words in a prompt can change the result significantly. Knowing them helps you get exactly what you want.

Word or phraseEffectExample
totalReturns a single figureTotal sales last month
by ...Breaks the result downSales by state last month
chartReturns a chart and tableChart sales by state last month
listReturns rows of detailList the 10 oldest inventory items
top / oldest / largestSorts and limitsTop 10 products by sales value
vs / compared toCompares two measures or periodsWalmart orders vs sales Q1 2026
email / printSends the resultEmail Angela Coles sales from Q1

Sharing prompts across your team

Once your team finds prompts that work well for your business, keep them. A shared list of the best prompts for each role, with your own customer, product and site names, is one of the fastest ways to bring new staff up to speed. Review the list every season, add new prompts that people discover, and remove ones that are no longer useful. When a prompt consistently gives the wrong answer, rate it and leave a comment so the Producepak team can train Pak-Bot to handle it.

Ready to try them? Read the getting started guide, or book a demo to test these prompts on data that looks like your operation.

See Pak-Bot answer questions about your own data

Book a demo and we will load sample data that looks like your operation, or connect your existing Producepak account.

Book a demo