AI Prompting Patterns

Prompt patterns to get the most out of AVA.

Prompting Patterns for AVA

For complex asks, all Gen AI tools like AVA can benefit from some extra structure in your prompting. These patterns are not necessary to use AVA effectively, but they can help structure your prompts to get more precise, useful, and actionable responses from AVA.

1. RISE (Request, Information, Simplify, Examples)

  • Request: Clearly state what you need.
  • Information: Provide any necessary background information.
  • Simplify: Break down complex requests into simpler, smaller tasks.
  • Examples: Include examples or templates to guide the desired output.

2. CLEAR (Context, Length, Expectation, Audience, Revision)

  • Context: Offer context around the task or question.
  • Length: Specify the desired length or detail of the response.
  • Expectation: Describe what a successful outcome looks like.
  • Audience: Identify who the document or information is for.
  • Revision: Request a specific number of revisions or iterations if needed.

3. SMART (Specific, Measurable, Actionable, Relevant, Time-bound)

  • Specific: Be specific about what you're asking for.
  • Measurable: Define how you will measure success or completion.
  • Actionable: Ensure the request is something the AI can act on.
  • Relevant: Make sure the request aligns with your current needs or goals.
  • Time-bound: If applicable, set a deadline for when you need the response.

4. GUIDE (Goal, Understand, Identify, Detail, Explain)

  • Goal: State the goal of your request.
  • Understand: Make sure the request is understandable and clear.
  • Identify: Identify any key points or criteria that need to be addressed.
  • Detail: Provide details that can help refine the AI's response.
  • Explain: If necessary, explain why you're making the request or why it's important.

Using Prompting Patterns in AVA

AVA Tasks is already broken up in to Roles, Actions and Goals to help your structure your prompt effectively. For AVA Chat and Knowledge Search, you can format your prompt in the chat bar, Workspace, or any other text editor. Using square brackets and labels is a good way to help AVA understand your pattern. An example of this could look like this:

[Request]
I need you to create a marketing email.
[Information]
Our company offers a new AI-driven logistics platform that optimizes routes, reduces shipping costs, and improves delivery times. The key features include real-time tracking, predictive analytics for demand planning, and customizable delivery options.
[Simplify]
The email should introduce our AI-driven logistics platform and encourage signing up for a free trial on our website.
[Examples]
Use a professional yet approachable tone, focusing on how our technology makes logistics seamless and more affordable.

Get started today!
Contact a member of the DataRM team.

Businesses know they need AI, but they don't know where to start. DataRM can help keep you ahead of the AI wave.