> ## Documentation Index
> Fetch the complete documentation index at: https://docs.omnifact.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Basic Prompt Structure

> Learn the foundation of effective prompting - how to write clear, specific, and complete instructions for Omnifact.

<Card title="Foundations of Good Prompting" horizontal>
  When you get the basics right, it becomes much easier to get the results you want from Omnifact.
</Card>

## The Golden Rule: Be Specific and Complete

**Be specific and use complete sentences**. Think of each prompt as giving directions to a capable colleague who doesn't know your context.

**Provide context when it matters.** Omnifact remembers your conversation as you chat, so you do not need to repeat background details in every follow-up message. When you start a **new chat**, switch to a very different topic, or need the AI to work from specific materials, include the necessary context in your prompt.

Omnifact can also carry persistent context for you:

* **[Personalization](/en/platform/core-features/account-settings/user-profile-and-personalization#personalization)** (Account Settings): Your role, working style, and response preferences apply in every conversation.
* **[Chat Instructions](/en/platform/core-features/spaces/customizing-spaces)** (in Spaces): Space administrators can set standing instructions that define the assistant's expertise and tone for that Space.

## What Makes a Good Prompt

Every effective prompt should include these three elements:

<CardGroup cols={3}>
  <Card title="What you want" icon="question">
    The specific task or question you need help with
  </Card>

  <Card title="Context you have" icon="info">
    Relevant background information
  </Card>

  <Card title="What you expect" icon="check">
    Format, length, or style requirements
  </Card>
</CardGroup>

A good prompt combines:

```
[TASK] + [CONTEXT] + [EXPECTATIONS] = Good Results
```

**Example:**

* **Task**: "Analyze this customer feedback"
* **Context**: "from our Q4 survey about our mobile app"
* **Expectations**: "and identify the top 3 improvement priorities"
* **Complete Prompt**: "Analyze this customer feedback from our Q4 survey about our mobile app and identify the top 3 improvement priorities."

## See the Difference

The more specific and descriptive you are, the better results you'll get.

<Tabs>
  <Tab title="Marketing">
    **Vague**: "Marketing ideas" <br />
    **Specific**: "Generate 5 social media post ideas for a B2B software company launching a new project management tool, targeting small business owners"
  </Tab>

  <Tab title="HR">
    **Vague**: "Job description help" <br />
    **Specific**: "Write a job description for a Customer Success Manager position, including required skills, responsibilities, and company culture fit for a 50-person SaaS startup"
  </Tab>

  <Tab title="Finance">
    **Vague**: "Budget stuff" <br />
    **Specific**: "Create a monthly expense tracking template for a small marketing department with categories for advertising, software tools, and events"
  </Tab>

  <Tab title="Customer Service">
    **Vague**: "Customer support ideas" <br />
    **Specific**: "Create a customer support script for handling product returns, including step-by-step instructions for the customer and a list of common questions and answers"
  </Tab>

  <Tab title="Sales">
    **Vague**: "Sales ideas" <br />
    **Specific**: "Create a sales email sequence for a SaaS company targeting small business owners, including a personalized outreach message, a value proposition, and a call to action"
  </Tab>

  <Tab title="Product Management">
    **Vague**: "Product ideas" <br />
    **Specific**: "Prioritize the feature backlog for our mobile app's Q3 release, focusing on user retention improvements and technical debt reduction, and estimate the impact of each item."
  </Tab>
</Tabs>

## Common Problems and Quick Fixes

<AccordionGroup>
  <Accordion title="Problem: Responses are too generic">
    **Solution:** Add specific details and constraints

    * Include your exact context (role, company size, industry)
    * Specify format and length requirements
    * Be clear about what to include or exclude
  </Accordion>

  <Accordion title="Problem: Omnifact seems confused">
    **Solution:** Check if you're assuming context

    * Are you in a new chat or on a new topic without enough background?
    * Would this make sense to someone new to your situation?
    * Add the missing context directly in your prompt, or set it once in Personalization or Space Chat Instructions
  </Accordion>

  <Accordion title="Problem: Wrong format or style">
    **Solution:** Be explicit about expectations

    * "Format as \[bullet points/table/email/report]"
    * "Use \[professional/casual/technical] tone"
    * "Length should be \[specific word count/number of items]"
  </Accordion>
</AccordionGroup>

<Tip>
  Read your prompt out loud. If you have to add "you know what I mean" at the end, it needs to be more specific!
</Tip>

## Other Beginner Guides

* Learn [Being Clear and Direct](/en/guides/prompt-engineering/beginner/clear-direct) to eliminate confusion and get exactly what you need
* Explore [Assigning Roles](/en/guides/prompt-engineering/beginner/roles) to transform Omnifact into domain experts
* Master [Controlling Output Format](/en/guides/prompt-engineering/beginner/format-style) to get responses in exactly the format you need