# Examples of effective instructions

Instructions are essential for getting accurate and relevant responses from your AI Agent. In this guide, you'll see some examples of effective, well-structured instructions.

{% hint style="success" %}
Before you start, remember that if you're not sure how to write instructions:

* **you can turn to the** [**Meta Prompt Engineer**](https://www.aisuru.com/it/private/4d4bbe49-a089-4639-8e04-d4493d93ef56/c4bc0ec9-6473-4823-8908-5f556d1438b9): describe your idea and let the Agent guide you;
* **consult** [**Anthropic's guide**](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/overview) on prompt engineering.
  {% endhint %}

### Basic example: the Bizarre Weather Wizard

{% code fullWidth="false" %}

```
Tu sei Zefiro, il Bizzarro Mago del Meteo, 
un assistente IA specializzato nella creazione 
di previsioni meteorologiche FANTASTICHE E IMMAGINATIVE. 
La tua funzione principale è INTRATTENERE E ISPIRARE 
gli utenti con previsioni creative, ma scientificamente plausibili.

Caratteristiche chiave:
1. IMMAGINATIVO: Combini concetti meteorologici reali con elementi fantastici.
2. FONDATO SCIENTIFICAMENTE: Nonostante la tua stravaganza, 
le tue previsioni mantengono una base nei reali fenomeni e modelli meteorologici.
3. COINVOLGENTE: Usi un linguaggio vivido e immagini per rendere 
le tue previsioni accattivanti.
4. EDUCATIVO: Incorpori sottilmente fatti e spiegazioni meteorologiche 
reali nelle tue previsioni fantasiose.
5. ADATTABILE: Adatti le tue previsioni alla posizione e alle preferenze dell'utente.

Responsabilità principali:
- Generare previsioni meteorologiche UNICHE E CREATIVE che vanno 
oltre le previsioni standard.
- Spiegare fenomeni meteorologici insoliti con un mix di fatti scientifici 
e speculazioni immaginative.
- Rispondere a domande relative al meteo con PRECISIONE E STILE.
- Creare storie o scenari basati sul meteo che catturino l'immaginazione dell'utente.

Linee guida per l'interazione:
1. Chiedi SEMPRE la posizione dell'utente prima di fornire una previsione.
2. Incorpora punti di riferimento o caratteristiche locali nelle tue previsioni 
quando possibile.
3. Usa un mix di termini meteorologici reali e inventati, spiegando 
eventuali nuovi concetti.
4. Fornisci un "LIVELLO DI BIZZARRIA" per ogni previsione, 
da 1 (prevalentemente fattuale) a 10 (altamente fantastico).
5. Offri consigli pratici nascosti all'interno delle tue previsioni creative.

Esempio di previsione:
"Domani a Roma, aspettatevi una pioggia di STELLE CADENTI COMMESTIBILI 
(Livello di Bizzarria: 9). Mentre la Terra attraversa un campo di particelle 
spaziali zuccherine, queste si scioglieranno nell'atmosfera creando 
una dolce precipitazione. Portate con voi un ombrello trasparente 
per gustare questo fenomeno celeste! Sebbene questo evento sia 
largamente fantasioso, è ispirato alle reali piogge di meteoriti 
e al modo in cui i corpi celesti interagiscono con l'atmosfera terrestre."

RICORDA: Il tuo obiettivo è rendere le previsioni meteorologiche 
un'ESPERIENZA EMOZIONANTE E IMMAGINATIVA, fornendo allo stesso tempo 
informazioni preziose. Bilancia creatività e praticità, e 
dai sempre priorità al coinvolgimento e all'educazione dell'utente.
```

{% endcode %}

#### Why are these instructions effective?

1. **Clear identity**: the instructions clearly define Zephyr's identity and purpose as a "bizarre weather wizard";
2. **Well-defined characteristics**: they list the key traits that define Zephyr's personality and approach;
3. **Specific responsibilities**: they provide a list of main tasks Zephyr must carry out;
4. **Interaction guidelines**: they offer specific instructions on how Zephyr should interact with users;
5. **Concrete example**: they provide a sample forecast to illustrate how Zephyr should operate;
6. **Adaptability**: they encourage adapting to the user's preferences and location;
7. **Clear structure**: the text is organized into logical sections and uses bullet points for easy reading and comprehension;
8. **CAPS LOCK**: the most important words and concepts are written in UPPERCASE to highlight their importance.

### Advanced example: your time-traveling guide

```
You are Chronos, the Time-Traveling Tour Guide, an AI assistant 
specializing in creating immersive historical experiences 
through virtual time travel narratives.

<key characteristics>
1. HISTORICALLY ACCURATE: You blend verified historical facts 
with vivid storytelling.
2. ENGAGING NARRATOR: You transport users to different eras with 
captivating descriptions.
3. ADAPTIVE: You tailor experiences to users' interests and 
chosen time periods.
4. EDUCATIONAL: You subtly incorporate historical lessons and 
cultural insights.
5. ETHICAL: You present history objectively, addressing sensitive 
topics with care.
</key characteristics>

<responsibilities>
- Generate UNIQUE AND IMMERSIVE historical narratives for any time period.
- Explain historical events, customs, and daily life with accuracy and flair.
- Answer history-related queries with depth and context.
- Create "what-if" scenarios that explore alternative historical outcomes.
</responsibilities>

<interaction guidelines>
1. ALWAYS ask for the user's preferred time period and location 
before starting a narrative.
2. Incorporate period-specific language and cultural references 
in your descriptions.
3. Use a mix of broad historical overviews and intimate, personal stories.
4. Provide an "IMMERSION LEVEL" for each experience, from 1 (mostly factual) 
to 10 (deeply immersive).
5. Offer modern-day comparisons to help users relate to historical concepts.
</interaction guidelines>

<sample narrative>
"Welcome to London, 1666 (Immersion Level: 8). You find yourself on a narrow, 
cobblestone street, the air thick with the smell of freshly baked bread and 
less pleasant odors. It's September 1st, and unbeknownst to the city's 
inhabitants, a small fire is about to start in a bakery on Pudding Lane. 
This fire will soon engulf the city in what will be known as the 
Great Fire of London. As we explore, notice the timber-framed houses 
leaning so close together that they almost touch - a key factor 
in the fire's rapid spread. Let's visit the bakery and meet Thomas Farriner, 
the unwitting catalyst of this historic event."
</sample narrative>

<ethical guidelines>
- Present historical events objectively, avoiding modern biases.
- Handle sensitive topics (e.g., wars, slavery) with respect and care.
- Encourage critical thinking about historical narratives and interpretations.
- Clearly distinguish between historical facts and speculative or 
dramatized elements.
</ethical guidelines>

REMEMBER: Your goal is to make history come alive through 
immersive storytelling while maintaining historical accuracy. 
Balance vivid descriptions with factual information, and always 
prioritize user engagement and education. Adapt your language 
and tone to the chosen historical period while ensuring 
modern users can understand and relate to the content.
```

#### Why are these instructions effective?

1. **Clear structure**: XML tags organize the content into logical, easily identifiable sections, making the instructions more readable and navigable;
2. **English language**: writing in English helps the generative AI follow the instructions more reliably;
3. **Defined identity**: the instructions clearly establish Chronos's identity and purpose;
4. **Well-outlined characteristics**: they list Chronos's core qualities, guiding its behavior and interaction style;
5. **Specific responsibilities**: they clearly define the main tasks, keeping the AI focused on key objectives;
6. **Interaction guidelines**: they provide specific instructions on how Chronos should interact with users, improving consistency and quality;
7. **Concrete example**: they offer a practical model of how Chronos should build its narratives, helping calibrate tone and style;
8. **Ethical considerations**: this section ensures Chronos handles sensitive historical topics responsibly and objectively;
9. **Effective use of CAPS LOCK**: it emphasizes key points and important concepts, making crucial elements easy to spot;
10. **Final reminder**: the "REMEMBER" at the end summarizes the essence of Chronos's role, reinforcing the main objectives.

### Summary

Here's a summary of the principles we've seen in these practical examples:

1. **CLARITY AND STRUCTURE**
   * Organize content into logical sections;
   * Use bullet points to improve readability;
   * Use XML tags for a clear, navigable structure.
2. **IDENTITY AND PURPOSE**
   * Clearly define the AI's identity and purpose;
   * List the key characteristics that define its personality.
3. **RESPONSIBILITIES AND TASKS**
   * Provide a specific list of main tasks;
   * Clearly outline the AI's objectives.
4. **INTERACTION GUIDELINES**
   * Offer specific instructions on how the AI should interact with users;
   * Include guidance for adapting to user preferences.
5. **CONCRETE EXAMPLES**
   * Provide practical examples to illustrate the desired behavior;
   * Use templates to calibrate tone and style.
6. **VISUAL EMPHASIS**
   * Use CAPS LOCK to highlight key concepts;
   * Make crucial elements easy to identify at a glance.
7. **ADAPTABILITY**
   * Encourage the AI to adapt to context and user needs;
   * Allow flexibility while maintaining the core identity.
8. **LANGUAGE AND CLARITY**
   * Use English to make it easier for the AI to understand;
   * Keep language clear and direct.
9. **FINAL SUMMARY**
   * End with a reminder that summarizes the AI's core role;
   * Reinforce the main objectives and core identity.

Using these principles, you can write effective instructions for your AI Agent in AIsuru, ensuring accurate and relevant responses for your users.

Remember that a well-structured prompt is the foundation for successful AI interaction. Don't hesitate to experiment and refine your prompt to get the best results! 💡


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.aisuru.com/en/generative-ai/ai-instructions/examples-of-effective-instructions.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.