Coding with AI for Non-Programmers
A beginner course for office workers, analysts and small-business owners who have never coded: what code is and what AI coding assistants get wrong, how to describe a task with inputs, outputs and examples, how to run a formula, a small Python script or a simple web page safely, how to check results with known data, and how to keep versions and follow company policy. Five modules, each with a worked example, and a 10-question final quiz. Vendor-neutral and based on cited official sources. Not an accredited qualification.
What you will learn
- Explain and apply: Explain what code is, what AI coding assistants do and the typical ways they get things wrong.
- Explain and apply: Describe a task to an assistant with inputs, outputs, rules, examples and small steps.
- Explain and apply: Run a formula, a small Python script or a simple web page safely and use error messages to get explanations.
- Explain and apply: Check results with known data and protect files and secrets before running or sharing anything.
- Explain and apply: Keep versions, recognise when to call a professional and respect licences, copyright and company policy.
Modules
- What code is and what AI coding assistants do (and get wrong)
- Describing what you want: inputs, outputs, examples, small steps
- Running it safely: formulas, a small script, a simple web page
- Checking the result: known data, backups and secrets
- From script to habit: versions, professionals, licences, policy
Course facts
- Level
- Foundation
- Estimated study time
- about 75 minutes
- Final assessment
- 10 questions · Pass mark 80%
- Available in
- English, Français, العربية
- Price
- Free. Optional verifiable certificate: $5
- Last updated
- Published by
- AMAADOR ACADEMY
Who this course is for
Office workers, analysts and small-business owners who have never programmed and want to use an AI assistant for spreadsheets, small scripts and simple web pages.
Prerequisites
No programming experience. You should be comfortable with a spreadsheet and with saving and copying files. Access to an AI assistant is useful but not required.
Course content
1. What code is and what AI coding assistants do (and get wrong)
Code is a precise list of instructions that a computer follows exactly, in order, with no common sense added. A spreadsheet formula is code, a short Python script is code, and the page you are reading is built from code. The few lines below store two numbers and print their product. Nothing is guessed: the computer multiplies 20 by 3 and shows 60, and if you had mistyped a name it would stop and complain rather than try to understand what you meant.
price = 20 quantity = 3 print(price * quantity)
An AI coding assistant is a language model that predicts text. Given a plain-language request, it writes code that looks like the code it saw in its training, and many assistants can also explain code, suggest fixes and rewrite what you paste. This lets someone who has never programmed produce a working formula or script in minutes. It does not mean the assistant understands your business, your files or your intent; it produces a plausible answer, and plausible is not the same as correct.
2. Describing what you want: inputs, outputs, examples, small steps
The quality of generated code depends mostly on the quality of the request. A useful request names three things: the input (what you start with, such as a file with columns Date, Customer and Amount), the output (what you want at the end, such as a table of monthly totals per customer) and the rules in between (ignore rows with no date, round to two decimals). If you leave these out, the assistant fills the gaps with guesses, and you will not see which parts were guessed.
An example is worth a paragraph of description. Paste three to five fake rows of the kind of data you have, then write exactly what the result should be for those rows. This gives the assistant a target and gives you a ready-made test afterwards. Say also what tools you have: Excel or a spreadsheet in a browser, Python installed or not, a company laptop with restrictions. A script for a tool you cannot run is no use, and naming your level, for example that you have never programmed, makes the explanation simpler.
3. Running it safely: formulas, a small script, a simple web page
Choose the lightest tool that does the job. A spreadsheet formula such as the first sample below sums column C only where column B says Paid; it changes nothing outside its own cell and you can see the result at once. If a formula can solve your problem, prefer it to a script. Ask the assistant to explain each part, test it on a few rows, and only then copy it down the column. A formula can still be wrong, for instance by using the wrong range, so check it with a total you can compute another way.
=SUMIF(B2:B100, "Paid", C2:C100)import csv total = 0 with open("sales_copy.csv", newline="") as f: for row in csv.DictReader(f): total += float(row["amount"]) print("Total:", total)<h1>Opening hours</h1> <p>Monday to Friday, 9:00 to 17:00</p>
A small Python script is the next step, as in the second sample, which reads a copy of a file and adds up one column. Python is described in the official tutorial. Always run a new script on a copy named clearly, never on your only version, and prefer scripts that read data and print or write a new file rather than overwrite the original. If your employer restricts software, ask IT before installing anything. A simple web page, as in the third sample, is a text file saved with an .html extension and opened in a browser, which is a safe way to see results; MDN offers free beginner lessons on HTML.
4. Checking the result: known data, backups and secrets
The most reliable check a non-programmer can do is to run the code on data whose answer you already know. Take five rows, compute the expected total by hand or with a calculator, and compare. The sample below does the same in code: each assert line states a fact that must be true, and the script prints a message only if all of them hold. Add edge cases that often break things: an empty row, a zero, a negative number, a very long text, a date in a different format. If the assistant wrote the code and the tests together, also read the tests, because a wrong test can approve wrong code.
def vat(net, rate): return round(net * rate, 2) assert vat(100, 0.2) == 20.0 assert vat(0, 0.2) == 0.0 assert vat(50, 0.2) < vat(100, 0.2) print("tests passed")
Never run code you do not understand on important files. Before running anything that changes, moves or deletes files, ask the assistant to list in plain words every file the code will read, write or delete, and to say what happens if it fails halfway. Keep a backup of the originals in a place the script cannot reach, and check that you can actually open the backup. Be especially careful with commands that mention deleting, formatting, recursive operations or administrator rights. If you cannot explain what a line does, the line does not belong in your working folder yet.
5. From script to habit: versions, professionals, licences, policy
A script that works today will be edited tomorrow, so keep versions. The simplest method is copies with dates in the name, such as sales_report_2026-11-20.py, kept next to a short note of what changed and what you tested. A better method is a version control tool such as Git, shown in the sample below, which records each saved state with a message and lets you go back. The Git documentation has a beginner book. Version copies also let you compare what the assistant changed when you asked for a small fix, which is a common source of surprises.
git init git add sales_report.py git commit -m "Add date filter, tested on 5 sample rows"
Know when to stop and call a professional. Do so when the work touches money movements, personal or health data, legal or regulatory reporting, customer-facing systems, or anything that many people will depend on daily. Also call for help when the script grows beyond what you can read in one sitting, when you keep fixing the same error, or when you are asked to connect to a company database or send emails automatically. A developer or the IT security team can review the code, apply company standards and maintain it when you are on leave. Ask for a review rather than asking permission to start: you bring the idea and the tested prototype.
The full lesson continues in the interactive course on this page.
Sources and references
- The Python Tutorial (official documentation)
- Learn web development (MDN)
- OWASP Top 10 for LLM Applications
- National Cyber Security Centre (official guidance)
- Guidelines for secure AI system development
- AI Risk Management Framework (AI RMF)
- Git documentation
- Copyright and Artificial Intelligence
Education only. It is not legal, financial or professional advice and not an accredited qualification. Official sources are summarised as read on the access date; check current terms, licences and your employer’s policy. Tools are named neutrally and none is recommended. Figures and scenarios in worked examples are illustrative.