8.14.0

Enhancements

• This release improves throughput performance by > 50% and also includes some latency improvements
• Improves AI Answer latencies to stream near instantly. Adds support for an AIConfig.strictSelection field for configurability of AI answer independent of search hits
• Improves tokens logic for AI answer to support a min and max token length boundary
• Adds a switch to disable logging with --disable-logs

Fixes

• Fixes caching behavior when there is an error in returning an AI answer response
• Sets analytics category for AI analytics endpoint, previous category was AI
• Fixes general use of pipeline's esQuery stage

8.14.0 RC

Enhancements

• This release contains performance improvements to rules, caching and efficient parsing through middlewares
• Adds a switch to disable logging, default behavior logs server request/responses
• Adds support for an AIConfig.strictSelection field for configurability of AI answer independent of search hits

Fixes

• Fixes caching behavior when there is an error in returning an AI answer response
• Sets analytics category for AI analytics endpoint, previous category was AI

8.13.4.beta

Changes in this release:

• Adds support for AIConfig.strictSelection
• Adds more debug data for the latency issue

8.13.4

Changes in this release:

• Adds more debug fields for the OpenAI latency issue
• Other fixes

8.13.3

• Adds support for capturing stats like when the OpenAI response got resolved, when SSE call gets resolved in order to improve OpenAI integration performance

8.13.2

Fixes

• This release standardizes the session creation for AI search request that's returned from cache

8.13.1

Enhancement

• Adds an endpoint to return an AI Answer directly, i.e. without passing through search
• Adds support for ordering of FAQs

Fixes

• Fixes a fatal error to be gracefully handled for syncing FAQs to Zinc
• Use precise error status codes for FAQs endpoints

8.13.0

Enhancements

General

• Enables AI search, pipelines, and search relevancy features for the 2023 (now default) cluster plans. This makes these features available to all new paid plan customers

Pipelines

• Optimizes latency of the boost stages for pipelines
• Optimizes pipeline invocation calls reporting to happen in batch instead of per each call
• Adds ?debug=true API flag support to see the response time per stage in the pipeline response

These enhancements make pipelines usable for production scenarios.

AI Search

• Improves AI Answer Dx by returning error when a user-specified maxToken value exceeds limit, informing the max valid value that can be set
• Adds initial support for FAQ type of suggestions: An AI or human-provided answer can be marked as FAQ and be displayed in the SearchBox UI to end-users
• Adds support for returning index and document_ids in the session details API endpoint

Build

• Uses Go1.20.4 as the build image for both Cloud and Docker images
• Includes build time profiling optimizations - expected ~2-4% performance improvement over 8.12 release

1.0.13

• Includes all changes in this release
• Adds support for allowing 10 pipelines if feature_pipelines is true
• Adds support for returning feature_openai in plan details
• Adds support for returning 402 when index count or storage is exceeded.

1.0.12

Changes in this release:

All the changes from this Arc release: https://github.com/appbaseio/reactivesearch-api/releases/tag/8.13.0