- Added `TURBOPUFFER_REGION` to the environment configuration.
- Modified the `createClient` function to utilize the new `TURBOPUFFER_REGION` variable.
- Updated the `generateNamespace` function to return the data source ID directly instead of prefixing it.
- Adjusted various tests and sync job logic to reflect changes in namespace generation and ensure consistency across the application.
Updated DEFAULT_CACHE_OPTIONS in analyst and think-and-prep agents to include gateway options. Added gateway settings in multiple steps including analysis-type-router, create-todos, extract-values, generate-chat-title, flag-chat, format-follow-up, format-initial, and identify-assumptions steps to restrict provider usage to 'anthropic'. This enhances control over the AI model interactions.
Update AI SDK dependencies and refactor model initialization to use AI Gateway. Remove deprecated fallback logic and related tests for improved performance and maintainability.
- Introduced the `processSyncJob` function to handle the complete workflow for syncing searchable values, including fetching credentials, querying distinct values, deduplication, generating embeddings, and upserting to Turbopuffer.
- Added support for the `@buster/search` package in the project configuration.
- Updated dependencies to version 4.0.2 for `@trigger.dev/build`, `@trigger.dev/core`, and `@trigger.dev/sdk`.
- Enhanced logging for better traceability during the sync process.