Get interaction details V2
Retrieve the detailed information about a single interaction using the v2 schema. This version includes expanded metadata such as sentiment, emotions, latency, PII info, and feedback details. You can list interactions using the Get interactions endpoint. For authentication you need the to use the API key associated to the project you want to retrieve the interactions from.
ID of the interaction you want to retrieve.
Response
Unique identifier of the interaction.
The project identifier to which the interaction belongs.
Sentiment classification of the interaction, if available (e.g., positive, neutral, negative).
Key-value pairs describing emotion probabilities or intensities detected in the interaction.
The cost incurred for the interaction
Automatically or manually assigned topic of the interaction.
The detected intent associated with the interaction.
The user input as received by the system.
The LLM-generated output corresponding to the input.
Time taken to generate the response.
Flag indicating whether PII (Personally Identifiable Information) was removed from the input.
Language detected in the user input.
The timestamp when the interaction began (user input was received).
The timestamp when the assistant’s response was generated.
User identifier associated with the interaction.
Number of characters in the output.
List of action labels inferred or triggered from the interaction
Additional context or metadata about the actions taken.
Actions taken or suggested via user feedback (e.g., “thumbs_up”, “thumbs_down”).
Text content accompanying feedback actions, if any.
Numerical rating provided as feedback (e.g., 1–5).
A normalized or cleaned-up version of the original user query, if processed.
Key-value pairs showing custom tags associated with the interaction.
Type of feedback provided (e.g., “negative_implicit”, “positive_implicit”).
A flag or description for any risky or concerning user behavior detected.
Identifiers or summaries of RAG sources used, if provided.
Detailed metadata about each RAG item (e.g., source, query, result).
Description of any warnings or issues detected with the LLM response.
List of conversation turns surrounding the interaction.
Get interaction details V2
Retrieve the detailed information about a single interaction using the v2 schema. This version includes expanded metadata such as sentiment, emotions, latency, PII info, and feedback details. You can list interactions using the Get interactions endpoint. For authentication you need the to use the API key associated to the project you want to retrieve the interactions from.
ID of the interaction you want to retrieve.
Response
Unique identifier of the interaction.
The project identifier to which the interaction belongs.
Sentiment classification of the interaction, if available (e.g., positive, neutral, negative).
Key-value pairs describing emotion probabilities or intensities detected in the interaction.
The cost incurred for the interaction
Automatically or manually assigned topic of the interaction.
The detected intent associated with the interaction.
The user input as received by the system.
The LLM-generated output corresponding to the input.
Time taken to generate the response.
Flag indicating whether PII (Personally Identifiable Information) was removed from the input.
Language detected in the user input.
The timestamp when the interaction began (user input was received).
The timestamp when the assistant’s response was generated.
User identifier associated with the interaction.
Number of characters in the output.
List of action labels inferred or triggered from the interaction
Additional context or metadata about the actions taken.
Actions taken or suggested via user feedback (e.g., “thumbs_up”, “thumbs_down”).
Text content accompanying feedback actions, if any.
Numerical rating provided as feedback (e.g., 1–5).
A normalized or cleaned-up version of the original user query, if processed.
Key-value pairs showing custom tags associated with the interaction.
Type of feedback provided (e.g., “negative_implicit”, “positive_implicit”).
A flag or description for any risky or concerning user behavior detected.
Identifiers or summaries of RAG sources used, if provided.
Detailed metadata about each RAG item (e.g., source, query, result).
Description of any warnings or issues detected with the LLM response.
List of conversation turns surrounding the interaction.