AI at work

AI for emails, spreadsheets, coding and images: where it saves time, where it fails and how to protect company data.

  • Intermediate
  • 15 min
  • 10 questions
  • Lesson 7 of 16

In this lesson you will learn

  • Judge the benefits and limits of AI for work tasks
  • Write prompts with placeholders instead of personal data
  • Check formulas and code produced by AI

Elvin works at a small furniture shop in Baku. This morning he has to answer twenty customer emails, prepare a sales report in Excel and make a poster for a weekend sale. An AI assistant can speed up every one of these tasks — if Elvin knows where it helps, where it fails and what he must never paste into a chat.

Where AI helps at work

TaskHow AI helpsWhat stays your job
Emails and textsDrafts, changes the tone, shortens, translatesCheck facts, names and tone — you are the sender
SpreadsheetsExplains and writes formulas, suggests chartsTest every formula on values you can check
CodingExplains code, finds bugs, writes small functionsRead, test and understand the code; check security
ImagesQuick drafts, illustrations, variationsCheck rights, the words in the image and company rules
Meetings and documentsSummaries, lists of decisions and tasksMake sure nothing important is lost or twisted

Sign up free to keep reading

The rest of the lesson, the quiz and your notes open with an Educora account. Signing up takes a minute, and your progress is saved on all your devices.

Writing: drafts with placeholders

AI is excellent at first drafts: a polite reply to a complaint, a product description, a short announcement. Give it the facts, the tone and the length. Where personal data would go, use placeholders such as [NAME] or [PHONE] and fill them in yourself later — the AI does not need the real data to write a good text.

Vague

write an email to a customer about a delay

Clear, with placeholders

You are a customer service manager at a
furniture shop in Baku. Write a short, polite
email to [NAME], whose sofa delivery moves
from 12 May to 15 May because of a supplier
problem.

- apologise once, do not over-explain
- offer free delivery as compensation
- end with: call us at [PHONE]
- at most 120 words, warm and professional
The strong prompt gives facts, tone and length and keeps personal data out: a person fills in [NAME] and [PHONE].

Spreadsheets: describe the table, test the formula

To get a formula, describe the structure of your table — which columns contain what, in which rows — instead of pasting real business data. Then test the formula on rows you can calculate in your head. AI sometimes mixes up ranges, uses a function your Excel version does not have, or uses the wrong separator (some language settings use a semicolon instead of a comma).

Example prompt: asking for a formula

I have an Excel table. Column A: product name, column B: price in manat, column C: quantity sold, rows 2 to 5.

1) Write a formula for column D that shows the revenue for each row.
2) Write one formula that adds up the revenue only for products that cost more than 100 manat.

Explain each formula in one sentence.

This prompt contains no real data — only the structure of the table. Check the formulas the AI suggested in the sheet below.

Interactive

Loading simulation…

Coding: an assistant, not an autopilot

Chat assistants and coding tools such as GitHub Copilot help programmers explain unfamiliar code, find bugs and write small functions and tests. Even a beginner can ask: “Explain this code line by line.” But generated code can contain bugs and security holes, and it often ignores unusual inputs. Use only code you understand, and never paste passwords or secret keys into a prompt.

The AI's first version

def average(numbers):
    return sum(numbers) / len(numbers)

print(average([4, 5, 6]))  # 5.0
print(average([]))         # ZeroDivisionError

After review and testing

def average(numbers):
    if not numbers:
        return 0
    return sum(numbers) / len(numbers)

print(average([4, 5, 6]))  # 5.0
print(average([]))         # 0
The first version crashes on an empty list. Testing edge cases — an empty list, zero, negative numbers — is the programmer's job.

Images: fast drafts with limits

Image generators create pictures from a text description: a poster idea, an illustration for a presentation, several design variants in minutes. Typical weaknesses: letters and words inside the image are often misspelled, and small details such as hands or logos can come out wrong. Rules on copyright and on using AI images differ between countries and companies, so check them — and never create realistic images of real people without their consent.

Practice: automate one routine task

  1. Choose a task

    Pick something you do every week: a report, standard replies, a table you clean by hand.

  2. Write the prompt

    Add role, context and format, and use placeholders instead of real personal data.

  3. Check the result

    Read every line, test formulas and code, compare with your own version.

  4. Measure

    Note how many minutes you saved and how many corrections you had to make.

  5. Write your checklist

    List what you always check for this task, so next time is faster and safer.

Remember

  • AI speeds up drafts, formulas, code and sketches, but checking and responsibility stay with you.
  • Use placeholders such as [NAME] and [PHONE] instead of personal data.
  • To get a formula, describe the table's structure, not its data, and test the formula on a row you can check in your head.
  • Test AI-written code on edge cases and use only code you understand.
  • Use only tools your organisation has approved for company data.
Spotted a mistake or something inaccurate in this lesson?

Check yourself

10 questions. Every correct answer earns XP.

Question 1 / 10

Question 1 / 10. Elvin wants an AI to draft a reply to a customer. How should he handle the customer's name and phone number?