Release April 2024

Overview

This release enhances the integration between the platform and AI backend services by enabling support for multiple provider IDs in various expert suggestion workflows. It improves the accuracy and relevance of expert recommendations by querying across multiple providers simultaneously. The update refines request handling, response aggregation, and data processing related to provider identifiers.

New Features

Introduced support for passing multiple provider IDs as arrays to AI backend endpoints, allowing multi-provider querying for expert and squad suggestions.
Aggregation and ranking logic added to combine and sort expert and ghost matches received from multiple providers, returning top results.

Improvements / Changes

Refactored getSuggestedExperts and getSuggestedExpertSquad functions to accept arrays of provider IDs instead of single IDs.
Updated AI backend query parameters to use providerIds (plural) with comma-separated values to reflect support for multiple providers.
Modified response handling to process and merge expert and ghost data from multiple provider-specific responses before selecting the best matches.
Adjusted internal helper methods to always operate over arrays of provider IDs for consistency and future extensibility.
Enhanced logging within expert suggestion workflows to output arrays of provider IDs and their corresponding results for improved traceability.
Standardized provider ID usage across code paths, replacing fallbacks on single provider IDs with explicit arrays derived from organizational or team configurations.

Bug Fixes

Fixed inconsistent behavior where only a single provider ID was considered, leading to incomplete expert suggestions when multiple provider contexts are present.

Technical / Developer Notes

AI backend endpoints now expect providerIds query parameters with multiple IDs joined by commas instead of a single providerId.
Expert and ghost data structures returned from AI services are now keyed by provider ID; internal logic merges these keyed results into unified sorted lists.
Downstream consumers of expert data should be aware that returned expert objects include a providerId field to indicate the source provider.
Existing calls passing single provider ID are internally wrapped into arrays to maintain backward compatibility.

Acknowledgements