The perfect prompt

A good prompt is the difference between a useful answer and hours of back-and-forth. On this page, we’ll show you why prompting is so important, what the basic principles are, and which ready-made templates you can copy and use right away.

Why is good prompting important?

AI models deliver exactly what you ask for—no more, no less. If you ask poorly, you get generic answers, have to follow up, and waste time. Good prompting is therefore not a gimmick, but one of the most important productivity skills when working with innoGPT.

• ⏱️ Time savings: Precise prompts lead directly to usable results without the need for repeated follow-ups.

• ✨ Better quality: Well-thought-out prompts generate more professional and relevant answers.

• 🔁 Consistency: Structured prompts deliver consistently good results—ideal for recurring tasks.

• 💸 Lower costs: Fewer follow-up requests mean fewer tokens and thus more efficient model usage.

• 🤝 Teamwork: You can save clear prompts in the prompt library and share them with colleagues.

Structured prompts with a single click

You don’t have to formulate every prompt yourself. In innoGPT, you’ll find the ✨ star icon in the input field. Clicking it automatically transforms your keywords into a structured, professional prompt—with role, context, task, and format.

Here’s how it works:

1. Type your rough idea into the input field (e.g., “Email to customers regarding delay”).

2. Click the ✨ star icon to the right of the input field.

3. innoGPT It generates a structured prompt from it in seconds.

4. Send the suggestion right away or tweak it first.

   

The basic principles

1. Clarity over brevity

AI models aren’t mind readers. The more specific your request, the more precise the answer.

❌ Vague:

​ Schreibe etwas über Marketing ​

✅ Precise:

​ Erstelle eine 200-Wort-Einführung über Content-Marketing-Strategien für B2B-Unternehmen, mit Fokus auf LinkedIn-Aktivitäten. Zielgruppe: Marketing-Manager mit 3–5 Jahren Erfahrung. Tonalität: professionell, aber zugänglich. ​

2. Context is king

The more relevant background information the AI receives, the more appropriate the answer will be. Explain who you are, who you’re writing for, and in what context the output is needed. Also use the voice feature to provide context more easily and quickly, rather than typing too much.

​ Ich bin Vertriebsleiter eines SaaS-Unternehmens mit 50 Mitarbeitern. Wir verkaufen eine Projektmanagement-Software an den Mittelstand. Schreibe eine Follow-up-E-Mail an einen Interessenten, mit dem ich letzte Woche eine Produkt-Demo hatte. Ziel: einen zweiten Termin mit dem CTO vereinbaren. ​

3. Assign a role

If you give the AI a specific role, it will respond from that perspective—using appropriate vocabulary and level of detail.

​ Du bist ein erfahrener Steuerberater mit Spezialisierung auf E-Commerce-Unternehmen. Erkläre mir verständlich, was die Kleinunternehmerregelung ist, wann sie sich lohnt und welche drei häufigsten Fehler ich bei der Anmeldung vermeiden sollte. ​

4. Define the format

Tell the AI explicitly what the result should look like—table, list, continuous text, JSON, word count, tone.

​ Erstelle eine Tabelle mit 5 Spalten: Tool-Name, Hauptfunktion, Preismodell, ideal für, Bewertung (1–5). Befülle sie mit 8 KI-Tools für Content-Erstellung. Sortiere nach Bewertung absteigend. ​

5. Provide examples (few-shot)

If you want a specific style or structure, show the AI 1–2 examples. This is often more effective than any description.

​Formulate LinkedIn posts in the following style:

Example 1: "Today I learned: 80% of all meetings could be an email. The other 20%? Need a clear agenda item."

Example 2: "The best marketing strategy? Build a product that people talk about voluntarily."

Example 3: “I’ve uploaded a similar social media post for you; use this structure as a guide…”

Now write 3 posts in the same style on the topic of "Remote Work." ​

Tips for even better prompts

• Iterate: The first prompt is rarely the best. Refine it step by step instead of starting over.

• Specify what you don’t want: “Avoid marketing clichés” or “No bullet points” guides the model specifically.

• Have the AI formulate prompts for you—write me a prompt for a social media assistant that understands and can create our posts; here’s an example post.

• Use separators: With ### or ---, you can clearly delineate sections.

• Let the AI think: “Think step by step” or “Justify your answer” significantly improves the quality.

• Save what works: Good prompts belong in your prompt library—that way, you’ll never have to rewrite them again.