All micro-internships
Micro-internship brief ≈4h in one sittingno-code

Design an AI assistant for a real person

Custom AI assistants (GPTs, Gems) are how professionals actually use AI now. Build one for a real person in your life — a shopkeeper, a teacher, a parent — test it against their real situations, and fix where it breaks.

AI LiteracyPrompt Engineering

Posted by The AI Internships

The work brief

  1. 01Pick a real person and one job they do repeatedly (replying to customers, planning lessons, writing bills).
  2. 02Write system instructions for an assistant that does that job (use Custom GPTs, Gemini Gems, or a saved system prompt).
  3. 03Test it on 5 real scenarios from that person’s life — real messages, real questions.
  4. 04Find at least 2 places it fails or sounds wrong, fix the instructions, and re-test.

What you’ll produce

4 deliverables

Submission standard

Submit who it’s for, your final instructions, all 5 test transcripts, and the 2 failures with your fixes. Generic personas ("a businessman") won’t pass — we want a real person’s real job. Share a public link to your Custom GPT/Gem (or a share link to one real test chat) so we can see it exists and behaves as described.

  • Who is this for, and what job does it do for them?

    Short answerRequired

  • Your final assistant instructions (system prompt)

    Written responseRequired

  • 5 test scenarios with the assistant’s replies

    Five test scenarios and replies, each reproducible via the assistant you share below.

    Written responseRequired

  • 2 failures you caught, and how you fixed the instructions

    Written responseRequired

You’ll complete these inside your private workspace.

What you must submit as proof

This brief requires evidence an AI can’t fabricate.

  • Public link to your Custom GPT/Gem, or a share link to one test chat

    A public link to the assistant you built, OR a ChatGPT/Gemini share link of one full real test conversation with it.

    Public linkRequired

Submissions without this evidence cannot be submitted.

Protect other people in your proof. Blur faces, names, phone numbers and email addresses before you upload, and refer to anyone you worked with by role or number ("Listener 1", "the stall owner"). Your proof is only ever used to check your work — it is never published, never appears on your certificate, and is never shown in your public portfolio.

How your work is evaluated

The passing benchmark is 70/100.

Fit to a real person

20%

The assistant is clearly shaped around a specific person’s actual job, not a generic role.

Testing rigor

20%

Five genuinely different, realistic scenarios were tested.

Failure analysis

20%

Real failures were found and the fixes clearly address them.

The assistant actually exists

40%

Auto-fail if there is no working link, or the linked assistant/chat doesn't match the described persona, instructions and transcripts (invented).

How we grade your AI usage

30% of your score

Using AI is the point — it’s the skill this certificate proves. You’ll answer three short questions about how you used it: what you asked, what was wrong with its first answer, and what you changed. Specific, honest answers score high. “I pasted the brief and submitted the answer” scores near zero.