Indicaciones
Las instrucciones (prompts) son las indicaciones que proporcionas a un modelo para guiar su comportamiento y generar respuestas significativas. Esta guía explica cómo funcionan las instrucciones, las mejores prácticas para redactar indicaciones eficaces y las técnicas para mejorar la calidad de las respuestas en diferentes tareas de IA.
What Is a Prompt?
A prompt is the input you send to a model. It can be a simple question, a detailed instruction, or a structured conversation that defines what the model should accomplish.
The quality of your prompt directly influences the quality of the generated response. Clear, specific, and well-structured prompts help the model better understand your intent and produce more reliable results.
Anatomy of a Prompt
Most effective prompts contain four key elements.
Component | Purpose | Example |
|---|---|---|
Task | What the model should do | Summarize the article |
Context | Background information | Target audience is beginners |
Constraints | Rules or limitations | Maximum 150 words |
Output Format | Expected response structure | Return Markdown with bullet points |
Including these elements makes prompts easier for the model to interpret and improves consistency across responses.
Prompt Structure
A structured prompt provides the model with enough information to understand both the objective and the expected result.
Basic Prompt Example
Simple prompts work well for straightforward questions but may produce varied responses depending on the complexity of the task.
Structured Prompt Example
Adding context, constraints, and formatting instructions helps the model generate more focused and predictable results.
Prompt Comparison
Weak Prompt | Improved Prompt |
|---|---|
Explain AI. | Explain artificial intelligence in simple language for high school students using one real-world example. |
Write an email. | Write a professional follow-up email thanking the client after a product demo. Keep it under 150 words. |
Create a blog. | Write a 1,000-word SEO blog about renewable energy with H2 headings, FAQs, and a friendly tone. |
Small improvements in wording can significantly enhance the quality and relevance of the generated output.
Prompting Best Practices
Clearly define the task.
Provide relevant context.
Be specific about the desired outcome.
Include formatting instructions when necessary.
State any important constraints such as length or tone.
Test multiple prompt variations to compare results.
Common Prompting Mistakes
Mistake | Why It Matters | Recommendation |
|---|---|---|
Being too vague | The model has to guess your intent. | Clearly describe the task. |
Missing context | Responses may not fit your use case. | Include background information. |
Multiple unrelated tasks | Output becomes inconsistent. | Break complex tasks into smaller prompts. |
No output format | Responses may vary in structure. | Specify Markdown, JSON, bullets, or tables when needed. |
Prompt Design Workflow
Following a consistent workflow makes prompts easier to maintain and improves response reliability across different applications.
Prompt Templates
Content Generation
Code Assistance
Data Analysis
Frequently Asked Questions
Next Step
Now that you understand how prompts influence model behavior, continue to Messages to learn how prompts are organized within conversations and how message roles affect AI responses.