JetBrains Console Help

AI activity and impact

The AI activity and impact analytics segment lets you understand how AI supports your teams in everyday coding. It provides insights into code suggestions generated and accepted through different AI tools and features, as well as metrics that help you estimate how AI is used in your organization.

IDE and plugin version requirements

Data required for this analytics segment is sent from the IDEs with AI Assistant and Junie plugins. The following versions of the IDEs and plugins are supported:

  • IDE version 2025.3.0 or later

  • AI Assistant version 2025.3.0 or later

  • Junie version 253.487.77 or later

Earlier versions of IDEs and plugins don't send data required for this analytics segment. In the case of partially meeting version requirements, AI activity and impact will not include data for all AI tools and features. For example, if the IDE and AI Assistant versions are up to date but the Junie version is older than the minimum requirement, Junie-related data will not be included in the analytics.

AI tools and features included in the metrics

The charts on this page provide data for a specific subset of AI tools and features that provide AI-generated code. The subset is divided into two categories, based on the type of interaction with the user:

  • Chat-based interactions: Data related to chat interactions with AI models and coding agents, such as lines of code generated and inserted by agents or the number of messages sent in chat modes. The metrics include data related to actions performed in the following tools:

    • Junie, when used in Code mode. Junie-related data is included in the metrics for both ways of using Junie: through the Junie tool window and through AI Assistant.

    • Claude Agent, a third-party coding agent by Anthropic integrated with AI Assistant.

    • AI Assistant, for messages sent in Chat mode that are included in the counts on the Messages sent to AI chats chart.

  • In-editor interactions: Code suggestions generated inside the editor by the following features:

    • Code completion: Cloud-based completion that can autocomplete single lines, blocks of code, and even entire functions in real time based on the project context.

    • Next edit suggestions: Prediction of the code segments to change next and suggestion of likely edits.

General activity and impact metrics

General activity and impact metrics provide insights into the overall effectiveness of AI tools and features. The currently available metric is AI code acceptance rate.

AI code acceptance rate represents the overall percentage of accepted suggestions out of the total number of generated lines of code for all AI tools and features included in the collection of analytics data.

This metric is an indication of the general quality, relevance, and trust in AI suggestions.

Specific metrics

Specific metrics let you get more details about the activity and impact of AI tools and features for specific types of interactions. For more information about each metric, see the following sections.

AI-generated code and acceptance rate

This section includes charts that show the acceptance rate of AI-generated code and the related sub-metrics that were used to calculate the acceptance rate, which serves as a measure of how good and relevant AI suggestions are.

AI generated code and acceptance rate chart

Code generated by chat-based agents

This chart shows data related to the additions and deletions of lines of code generated by AI agents in chat interactions and the number of lines of code that the users accepted. The chart shows the following sub-metrics:

  • Lines of code rejected: The suggested lines of code that were discarded or rolled back.

  • Lines of code accepted: All added or removed lines of code that the users did not discard or roll back.

The data also includes the following sub-metrics:

  • Total lines of code generated: All generated lines of code resulting from agent runs. Does not include code generated in read-only modes, such as Ask mode in Junie or Chat mode in AI Assistant. For more information about available modes and what they represent, see Junie and AI Assistant documentation.

  • Acceptance rate: The percentage of accepted lines out of the generated lines of code.

This is what the code generation and acceptance flow looks like with different chat-based coding agents.

After a task run, Junie makes changes to one or more files and lists the changes along with the number of added and deleted lines of code per file. The sum of added and deleted lines represents the number of generated lines of code. This sum also represents the number of accepted lines of code before any rollbacks.

If the user decides to roll back some or all changes, the number of added and deleted lines that were rolled back is then subtracted from the total number of accepted lines of code. The following user actions in the IDE count as rollbacks:

  1. Clicking the Rollback All button at the beginning of a Junie task.

  2. Clicking the rollback button for changes made to a specific file.

  3. Clicking the Rollback button at the end of a task.

Junie-generated code

When using Claude Agent, a completed agent run results in a summary of changes to the files, along with the number of added and deleted lines of code per file. The sum of added and deleted lines counts as the number of generated lines of code. The user can then decide to accept or discard the changes made by the agent completely or partially. All accepted lines of code are added to the total number of accepted lines. Rejected lines of code are those that weren't accepted by the user through one of the following actions:

  1. Clicking the Discard all button. This discards all changes generated by the agent.

  2. Clicking Discard icon Discard for a single file. This action discards only the changes made to the specific file.

Code generated by AI Assistant

Code generated by in-editor features

This chart includes data related to the lines of code generated by Code completion and Next edit suggestions and the number of lines that the users accepted. The chart shows the following sub-metrics:

  • Lines of code rejected: The suggested lines of code that weren't accepted by the user.

  • Lines of code accepted: Depending on the feature, accepted lines of code are counted as follows:

    • Code completion: All lines of code that were inserted into the editor. When the user partially accepts a suggestion, lines are counted as follows:

      • When the user accepts a part of a single line, this counts as one accepted line.

      • In multi-line suggestions, the number of accepted lines is the number of lines that the user accepted to insert.

    • Next edit suggestions: The number of lines that the user accepted to insert into the editor or delete from the editor. In a multistep suggestion flow, the number of accepted lines is added to the total count after each step.

    For more information about user actions in the IDE to accept or reject suggestions, see Code completion and Next edit suggestions.

The data also includes the following sub-metrics:

  • Total lines of code generated: Depending on the specific code suggestion feature, generated lines of code are counted as follows:

    • Code completion: The total number of lines that were shown to the users as suggestions in the editor.

    • Next edit suggestions: The total number of shown lines in the editor that the user can choose to insert and delete.

  • Acceptance rate: The percentage of accepted lines of code out of all generated lines.

AI-modified code

This section includes charts that show the number of lines of code that were modified by AI tools and features.

Lines of code modified by AI

Lines of code modified by chat-based agents

This chart shows the number of lines that were added, modified, or deleted as a result of chat interactions with a chat-based agent. You can optionally select to show data only for a subset of agents.

Lines of code modified by in-editor features

This chart shows the number of lines that were added, modified, or deleted by in-editor code suggestion features. You can optionally choose to show data for a subset of features.

AI feature activity

This section includes charts that provide insights into the number of interactions with AI tools and in-editor features. For AI tools, interactions are measured as the number of messages sent through AI chat. For in-editor features, interactions are measured as the number of suggestions shown to the user.

AI feature activity charts

Messages sent to AI chats

This chart shows the number of messages that the users sent to AI chats for the selected period. It provides insights into the number of interactions with AI tools over a period of time, which can help you understand agent adoption and usage volume trends in your organization.

Each message that the user sends increases the number of sent messages by one.

You can optionally select to show data only for the selected agents.

Suggestions shown by in-editor features

This chart shows the number of suggestions that the in-editor code suggestion features showed to the users. Each time the editor shows a suggestion increases the number of suggestions shown by one, regardless of how many lines the suggestion includes. You can optionally select to show data only for the selected AI features.

08 December 2025