Prompt engineering in your work and career
Turn prompt skills into value: what clients buy, roles that use the skill, a portfolio method, scoping and pricing a small job, and honest practice.
In this lesson you will
- Explain what clients value
- Build a portfolio
- Scope and price a small job
- Work honestly with clients
About 18 minutes
Turning a skill into something people value
Imran is good at prompts. His cousin’s company wastes hours writing customer replies, and Imran builds them a set of five tested templates. They are delighted, and his cousin asks, ‘What do I owe you?’ Imran has no idea. He undercharges, spends two unpaid weeks on revisions, and learns the lesson of this lesson: a skill becomes a livelihood only when you can explain what you deliver, show evidence, set the scope and price it fairly.
This lesson is about method. It does not promise jobs, income or demand. Markets, pay and job titles differ by country and change over time. Check current listings and rules where you live.
What employers and clients actually buy
| They do not buy | They buy |
|---|---|
| ‘I know prompts’ | Replies drafted in minutes that staff only need to review |
| A clever prompt | A tested template with rules, examples and a test set |
| An AI chat | A repeatable process with checks and fallbacks |
| Speed alone | Speed plus quality plus fewer mistakes |
Roles where prompt skills help
- Support and operations: reply drafts, triage and summaries.
- Marketing and content: briefs, outlines and variations that people edit.
- Education: lesson materials, quizzes and feedback drafts.
- Data and admin work: extraction, cleaning and classification with checks.
- Product and software teams: prompts inside apps, test sets and evaluation.
- Freelance: small packages for local businesses, creators and teachers.
Few employers hire for ‘prompt engineer’ alone. The skill more often adds value to a role you already do or want. Use the job’s own words in your applications, such as ‘built a reply system’, ‘cut drafting time’, or ‘created a quality checklist’.
Building a portfolio
- Choose three real problems (even small, even your own).
- Document each one: the problem, the prompt template, the test set, the results and what you changed.
- Show evidence honestly. Use before-and-after samples, label them as examples, and remove private data.
- Add a short README so a stranger can use the template.
- Publish where you are comfortable, such as a simple web page or a shared document.
Scoping and pricing a small job
Clients value clarity. Write down what you will deliver (number of templates, test cases, a short guide), what is not included, how many revision rounds are included and how long it takes. Then price it. A simple method is to estimate hours for each part, add testing and revisions, and multiply by an hourly rate you choose from your market and your experience. This program does exactly that, with example numbers only. The output is real arithmetic:
# A transparent quote calculator for a prompt-writing job (all numbers are examples, not market rates)
def quote(hours_per_template, templates, test_hours, revision_rounds, hourly_rate, support_days_fee=0):
build = hours_per_template * templates
testing = test_hours
revisions = revision_rounds * 1.5 # assume 1.5 hours per revision round
total_hours = build + testing + revisions
price = total_hours * hourly_rate + support_days_fee
return total_hours, price
rate = 20 # example hourly rate; choose your own from your market and skills
hours, price = quote(hours_per_template=2, templates=5, test_hours=3, revision_rounds=2, hourly_rate=rate)
print(f"Estimated hours: {hours}")
print(f"Price at {rate} per hour: {price}")
hours2, price2 = quote(2, 5, 3, 2, rate, support_days_fee=40)
print(f"With a paid 14-day support fee: {price2}")
Estimated hours: 16.0
Price at 20 per hour: 320.0
With a paid 14-day support fee: 360.0
The rate of 20 is just an example. Look at what similar services charge where you are, think about your costs and taxes, and adjust. Quote a clear price for a clear scope, and charge extra for work beyond it.
Honest practice with clients
- Say that outputs need human review, and never promise perfect accuracy.
- Never present AI work as something it is not, and follow your client’s and platform’s rules on disclosing AI use.
- Do not put clients’ private data into tools they have not approved.
- Keep a written agreement, even a short one, covering scope, price, ownership of the work and confidentiality.
- Be careful with any ‘get rich with prompts’ offers. If a course or job promises easy income, check carefully before paying.
Keeping up
Tools change quickly, but the method does not: goals, context, examples, formats, checks, tests. Keep a learning log, re-test your templates when tools update and read each tool’s current guide. Teaching someone else what you know is also a strong way to keep it sharp.
Practice
Write a one-page offer for a small business you know: the problem, what you would deliver, what is excluded, how many revisions, how long it takes, and how you would price it using hours. Then write three sentences describing the work in resume language, without exaggeration.
A good scope has numbers and limits: ‘five templates, ten test cases, one short guide, two revision rounds, delivered in ten working days’. A good resume line is specific and honest: ‘Built and tested five reply templates for a ten-person shop; drafting time per reply fell in our trial’. Use only results that are true.
Pricing by guessing and leaving scope open. Without limits, unpaid revisions and extra requests eat the whole fee.
Check what you learned
Five quick questions. Pick an answer and you will see straight away whether it is right and why.
What do clients really buy?
- The word prompt
- A long chat
- A password
- Tested, repeatable results
Outcomes and reliability.
What belongs in a portfolio item?
- Only a screenshot
- Problem, template, tests, results, changes
- Private client data
- A promise
Evidence a stranger can follow.
Why define scope?
- To hide prices
- To avoid agreements
- To skip tests
- To limit revisions and extras
Clear limits protect both sides.
How should you treat AI outputs with clients?
- Say they need human review
- Promise they are perfect
- Hide them
- Never test
Honesty builds trust.
Where should you get your hourly rate from?
- A random number
- An online promise
- Your market, costs and experience
- A friend’s guess only
The example rate is only an example.
Summary
- Sell outcomes: tested, repeatable, checked.
- Document three real problems with tests and results.
- Write the scope, revisions and price down.
- No income promises; be honest about AI use and review.
Questions
Is 'prompt engineer' a real job?
Some listings use the title, but the skill more often appears inside other roles. Check current job listings where you live.
How much should I charge?
There is no single answer. Estimate your hours, check what similar services charge locally, and adjust for your costs and experience.
Do I need to tell clients I used AI?
Follow your client's and platform's rules, and be honest. When in doubt, say so in the agreement.
