Skip to main content

How to use prompts effectively in Jaicob

Learn how to write effective prompts in Jaicob to get better results from AI and work faster with less effort.

Updated over a month ago

What Are Prompts in Jaicob?

Prompts are short instructions you give to the AI to perform specific tasks, such as sourcing candidates, generating outreach messages, summarising CVs, generating reports, send meeting invites and much more.

They are your way of communicating what you need and the more clear and specific you are, the better the results.


Using the Copilot for Prompts

In Jaicob, all prompts are handled via the Copilot, located at the top left of your screen.

You can open the Copilot from any page (except Settings) and type a request—no need to switch sections.

Whether you’re viewing a candidate, job, or agent, you can prompt actions directly. The Copilot works across the entire platform and understands context.

Examples:

  • “Send the profile of @Peter Smith to [email protected]

  • “Find candidates for Marketing Manager Amsterdam”

Let the Copilot handle the task for you.


Writing a Good Prompt

A good prompt is clear, specific, and has context. Here’s a simple structure to follow:

1. Define the goal – What do you want the AI to do?

2. Give context – Who is the candidate, client or job?

3. Specify format (if needed) – Do you want a bullet list, short paragraph, or email style?


Examples

✅ Good Prompt:

Write a short outreach message (max 5 lines) for a senior .NET developer in Berlin who recently worked at a fintech company. Focus on growth opportunities.

❌ Poor Prompt:

Write a message for a developer.

✅ Good Prompt:

Summarise this candidate’s experience based on the job description for a Medior Java role at a scale-up. Use 3 bullet points.

❌ Poor Prompt:

Give me a summary.


Tips for Better Results

  • Be specific: include seniority, location, or industry if relevant

  • Keep it short and clear, long prompts can confuse the AI

  • If something doesn’t work, tweak the wording slightly

Did this answer your question?