Prompt a Natural Language Aggregation

On this page

About this Task

Before you Begin
Steps
Example
Next Steps
Learn More

You can use MongoDB Compass to generate aggregation queries using natural language. Compass uses AI to generate aggregations based on prompts you provide. Querying with natural language can be a helpful starting point and assist you in learning to write MongoDB queries.

About this Task

Note

When you query your data using natural language in Compass, the text of your prompts and details about your MongoDB schemas are sent to Microsoft and OpenAI for processing. Your data is not stored on any third party storage systems or used to train AI models.

This software uses generative artificial intelligence. It is experimental and may give inaccurate results. Your use of this software is subject to MongoDB's:

You can also provide natural language prompts on the documents tab.
This page uses the sample_mflix.movies collection from the Atlas sample dataset.
It can be difficult to predict the complexity of the query or pipeline returned by the Compass AI. You may need to adjust the MAX TIME MS query bar option to either allow longer query processing time or prevent long-running queries. For details, see Adjust Maximum Time for Query Operations.

Before you Begin

You must enable natural language querying in Compass.

Steps

Navigate to the Natural Language Query Bar

Select the Aggregations tab.
Click the Generate aggregation button.

The Natural Language Query Bar displays.

Type a question about your collection

Type a natural language prompt for your collection into the query bar. Aggregation pipeline prompts usually have an aggregation verb such as count, average, or sum with logical conditions. For example: How many movies have more than 3 writers in the writers array?

Press enter or click the Generate aggregation button.
An aggregation pipeline populates in the Pipeline bar. You can scroll down to see the syntax of each stage.

Run the aggregation

Before running the query, make sure to thoroughly review the syntax of each stage. Ensure the returned pipeline has the fields and stages that match your desired use case.
Warning
Some aggregation operators, like $merge and $out, can modify your collection's data.
If your aggregation pipeline contains operators that can modify your collection's data, you are prompted for confirmation before the pipeline is executed.
Press enter or click the Run button to execute the pipeline.

The results populate in the aggregations view.

Tip

You can optionally provide feedback by clicking the or icon by the Natural Language Query Bar and provide details on your experience.

Your feedback is not used to train any AI models.

Example

Below are examples of prompts to help you understand expected results when using natural language prompts for aggregation.

Prompt

Response

Count the movies that have a release year of 1999

[
   {
      $match: {
         year: 1999
      },
   },
   {
      $count: "total_movies",
   }
]

Which comedy genre movie has the most awards?

[
   {
      $match: {
         genres: "Comedy"
      }
   },
   {
      $sort: {
         "awards.wins": -1,
         "awards.nominations": -1
      }
   },
   {
      $limit: 1
   },
   {
      $project: {
         _id: 0,
         title: 1,
         "awards.wins": 1,
         "awards.nominations": 1
      }
   }
]

How many movies have a imdb.rating > 4?

[
   {
      $match: {
         "imdb.rating": { $gt: 4 }
      }
   },
   {
      $group: {
         _id: null,
         count: { $sum: 1 }
      }
   }
]

Next Steps

Prompt a Natural Language Query

Learn More

Query with Natural Language

Back

Prompt Query

AI & Data Usage