All Collections
For Users:
Proficient users looking to get more out of Grain
Ask anything of Grain’s AI about your recordings
Ask anything of Grain’s AI about your recordings

Learn how to ask Grain AI anything about a particular meeting recording and add it to your notes

Updated over a week ago

Grain allows you to extract important information you need from your meetings using Ask Anything. However, the quality and usefulness of the information you receive can depend on the prompt you use.

Here we'll share the best practices to help you write more effective prompts, enabling you to get the best information possible!

Ask Grain AI anything

You can navigate to the 'Ask Anything' button by clicking on a specific meeting and selecting the 'Notes' tab.

Then, a pop-up will appear allowing you to type in your prompt

Tips to writing effective prompts

Keep it simple and precise

When creating prompts for Grain, simplicity and precision are your allies. Imagine you're explaining the task to a fourth-grader: straightforward and clear. State the specific task or analysis you want Grain to perform on your meeting. For instance, if you desire a concise summary or highlights from a meeting, ask the prompt to "provide key takeaways," "highlight notable events," or "provide sentiment analysis." This approach not only streamlines the process but also ensures you get precisely what you need.

Bad Example: "What was discussed in the meeting?"

Good Example: "Please provide a concise summary of the key points discussed in the meeting."

Request specific responses

Sometimes, you need a straightforward 'Yes' or 'No' answer to a specific question. If that's the case, start with clear phrasing. For example, you can say, "Please respond with 'Yes' or 'No.' Did this call include a discussion about pricing?" This approach not only narrows down the AI's response options but also guides the AI to look for the exact information you're seeking.

Bad Example: "Tell me about the pricing."

Good Example: "Please respond with 'Yes' or 'No.' Was a pricing discount part of this conversation?"

Use plain language, avoid jargon

In your prompts, it's essential to communicate with simple words and avoid jargon. The goal is for your prompts to be easily understood by anyone who reads them. This clarity ensures that the AI can interpret your prompts accurately. For instance, instead of saying "Please elucidate the deliverables," you could say, "Can you explain what needs to be delivered?" This not only simplifies the communication but also enhances the chances of receiving the desired response.

Bad Example: "Please elucidate the deliverables."

Good Example: "Can you explain what needs to be delivered?"

Mention key features

If you're interested in specific elements within the meeting, such as identifying specific tasks, action items, or notable events, it's a good practice to mention them explicitly in your prompts. For example, you can say, "Please identify any CRM items discussed during the meeting." By focusing on the specific elements you're interested in, you guide the AI's attention and improve the relevance of the responses.

Bad Example: "Tell me what's important."

Good Example: "Please identify any CRM items discussed during the meeting."

Consider the response format

Depending on your needs, it can be beneficial to specify the format in which you want the analysis presented. This consideration is particularly important if you have a preferred way of consuming information. For instance, you can request the summary in bullet point format if that's how you prefer to review the insights. By specifying the response format, you streamline the presentation of insights to align with your preferences.

Bad Example: "Give me the results."

Good Example: "Please provide a summary in bullet point format."

Focus on what you want the prompt to achieve

Instead of concentrating on what you don't want your prompts to do, be specific about what you want them to accomplish. For instance, use "Please summarize the following call in 3 sentences or less" rather than "Please summarize the following call. Don't write more than 3 sentences."

Bad Example: "Please summarize the following call. Don't write more than 3 sentences."

Good Example: "Can you provide a concise summary of the call in 3 sentences or less?"

Start with action verbs

Begin your prompts with action verbs to clearly define what you want the AI to do. Action verbs provide a direct and unambiguous instruction, helping the AI understand your request.

Bad Example: "Discuss the project progress."

Good Example: "Summarize the project progress."

Use keywords and synonyms

Incorporate relevant keywords and synonyms into your prompts. This can help capture a wider range of information and ensure that the AI doesn't miss any important details. For example, if you're interested in information about customer satisfaction, you can include synonyms like "client happiness" or "user contentment" in your prompts.

Bad Example: "Tell me about customer satisfaction."

Good Example: "Provide insights on customer satisfaction, including client happiness and user contentment."

Use conditional statements

Introduce conditional statements in your prompts to handle various scenarios. This can be helpful when you expect multiple outcomes based on the data available.

Bad Example: "Summarize the budget meeting."

Good Example: "If the client mentioned budget constraints, provide suggestions for cost optimization. If not, summarize the key points discussed in the budget meeting."

Provide context

Sometimes, providing context for your prompts can be highly valuable. Briefly explain the background or the reason behind your request. This helps the AI understand the purpose of your query, especially when the context is critical for the analysis.

Bad Example: "Summarize customer feedback."

Good Example: "Given the recent product launch, summarize customer feedback on the new features discussed in the product meeting."

Iterate and refine prompts

Recognize that crafting the perfect prompt may require some trial and error. Don't hesitate to iterate and refine your prompts based on the results you receive. If a particular prompt isn't delivering the desired insights, adjust it and test it again. The process of refining your prompts can lead to more accurate and effective results over time.

Bad Example: "Summarize the project progress."

Good Example: "If "Summarize the project progress" isn't yielding the insights you need, you can refine it to say, "Summarize the key achievements and challenges in the project progress report."

Did this answer your question?