FAQ: API & Analytics Capabilities – AIforService (XO)

Knowledge Articles
, ,
1

:receipt: Overview

This article outlines responses to frequently asked questions related to API performance, analytics capabilities, and data integration on the Kore.ai platform.

These insights are particularly useful for technical teams, product managers, and enterprise clients seeking clarity on how metrics and custom data can be retrieved and utilized.

:pushpin: Frequently Asked Questions

1. Is there any delay in API responses? If yes, what is the expected value?

Yes, some APIs that return large datasets are handled asynchronously. In these cases:

  • An initial request returns a request ID
  • This ID can be used to poll the status
  • This approach is mainly applicable to Analytics APIs and is not common across all public APIs

:warning: Note: Most public APIs are synchronous with standard response times unless specified otherwise.

2. How long after a user session ends are metrics available?

For customers hosted on US SaaS, metrics typically become available within:

  • ~15 minutes post-session
  • In many instances, data may appear sooner than 15 minutes

This applies to analytics data related to conversations, user sessions, and performance.

3. Can we make an API call for an individual user’s metrics?

Yes. Some public APIs provide filtering options that allow you to query:

  • User-specific data
  • Session-level metrics
  • Tag or intent-based metrics

:white_check_mark: Ensure the correct filter parameters are applied to retrieve user-specific information.

4. Can Kore integrate the Feedback API with the Analytics API to return both feedback and analytics data in a single response?

Currently, this is not supported.

  • Feedback data and analytics data must be retrieved separately
  • No unified endpoint is available to merge these responses

:hammer_and_wrench: This integration may be considered for future platform enhancements.

5. Is there an ability to pass custom data in the response?

  • During a conversation: Yes, custom data (e.g., tags, metadata) can be passed using scripting and platform variables.

  • During data retrieval via API: You cannot dynamically pass custom data in API queries. However:

    • Previously stored custom data (like tags) can be retrieved via relevant Analytics APIs

6. Can we increase the return record limit per call (currently 50)?

  • On SaaS environments, this limit is currently fixed
  • On On-Prem deployments, it is configurable and can be increased based on tenant-specific needs

:pushpin: For high-volume data needs, batching or paginated queries are recommended.

7. Which API can be used for real-time data streaming?

The platform does not currently support real-time data streaming via APIs. For example:

  • Analytics events are not emitted in real time
  • Attempting real-time access without understanding rate limits may lead to throttling or blocked requests

:question: If your use case requires near real-time access, please consult the Kore.ai team to evaluate feasibility.

:brain: Summary Table

Query Supported? Notes
API delay for large responses :heavy_check_mark: (Async) Request ID is used for polling
Post-session metric availability :heavy_check_mark: ~15 min for US SaaS
Per-user metric API :heavy_check_mark: Use filters
Feedback + Analytics in one API :x: Separate calls needed
Pass custom data :heavy_minus_sign: Possible during conversation; retrievable post-session
Increase API return limit :heavy_minus_sign: SaaS: :x:, On-Prem: :heavy_check_mark:
Real-time streaming API :x: Not currently advised

:toolbox: Additional Resources