Algolia
Developer ToolsKeep Algolia search indices fresh with AI agents. Agents add, update, and delete objects, run searches, execute batch operations and delete-by-query cleanups, and manage index settings so results always reflect live data.
What This Integration Enables
Search is the front door of the product, and a search index is only as good as its freshness. A stale index sells products you no longer have, at prices you no longer charge. FlowRunner agents close that gap by treating the Algolia index as a continuously synced projection of the source of truth: Add Object and Save Object for new and replaced records, Partial Update Object for incremental changes like price and stock, and Batch Operations when whole feeds move at once. Every write is asynchronous on Algolia's side and returns a taskID, which agents carry through the flow to confirm completion before declaring the sync done. The other half of index hygiene is removal, and that is where this connector is deliberately careful. Delete By Query, Clear Index, and Delete Index are all irreversible for the records they remove, so agents pair them with evidence: search first, count what matches, and put a person in front of the result before anything is destroyed.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Storefront Search That Tracks the Catalog
When a product changes in [Shopify](/integrations/shopify), the agent writes it through Save Object under its stable objectID, or applies just the moved fields with Partial Update Object. New products are searchable within the sync cycle, and price changes reach search at the same time they reach the product page. Search Index runs as a post-write check, confirming the record surfaces for its primary query terms before the sync reports success.
Feed Import at Batch Scale
A supplier feed arrives as rows in [Google Sheets](/integrations/google-sheets). The agent maps each row to a record and loads the whole set with Batch Operations, mixing adds, updates, and deletes in one high-throughput request. The result summary, including counts and the returned objectIDs, posts to [Slack](/integrations/slack), and any rows that failed mapping are listed for a person to fix at the source rather than patched silently in the index.
Retiring a Product Line, With a Preview
A brand is discontinued and its records need to leave the index. The agent builds the structured filter, runs Search Index with the same filter to enumerate exactly what would be removed, and presents the count and sample records for approval. Only after sign-off does Delete By Query run. Index settings changes follow the same pattern: Get Settings captures the current configuration before Set Settings changes ranking or faceting, so every configuration change has a documented rollback.
Human-in-Loop Highlight
Delete By Query removes every record matching a filter expression, asynchronously and permanently, and its bigger siblings go further: Clear Index empties an index irreversibly, and Delete Index destroys the index with its settings, synonyms, and rules. So the cleanup agent never runs any of them on its own reading of a filter. It previews first: "This filter (brand: Northline AND status: discontinued) matches 412 records in the products index. Five samples attached. Deletion cannot be undone. Approve the delete, tighten the filter, or cancel?" The catalog owner answers, and the agent executes exactly the approved scope, then tracks the taskID to completion. That [human-in-the-loop](/concepts/human-in-the-loop/) preview is the difference between index hygiene and explaining to the team why search returns nothing.
Agent Capabilities
13 actionsRecords
5- Add Object Adds a record with an auto-generated objectID, for brand-new records without a known identity. Returns a taskID for tracking the asynchronous index.
- Save Object Creates or fully replaces a record at a specific objectID. The idempotent write that sync loops are built on.
- Partial Update Object Updates only the specified attributes, with built-in increment, decrement, add, and remove operations, and a Create If Not Exists switch. The right call for price and stock changes.
- Get Object Retrieves a record by objectID, optionally limited to selected attributes. The pre-write read that makes conflicts visible.
- Delete Object Deletes one record by objectID, asynchronously with a taskID. The narrow, targeted form of removal.
Search
1- Search Index Searches an index and returns ranked, paginated hits with highlighting, faceting, and filtering. Doubles as the verification step after writes and the preview step before deletes.
Bulk Operations
3- Batch Operations Executes many indexing operations against one index in a single request, mixing adds, updates, partial updates, and deletes. The high-throughput path for feed imports.
- Delete By Query Deletes all records matching structured filters (facet, numeric, or tag). Irreversible for the matched records, which is why it runs only after a previewed, approved scope.
- Clear Index Removes every record while preserving the index's settings, synonyms, and rules. Irreversible for the records; gated accordingly.
Index Management
4- List Indices Lists the application's indices with record counts, sizes, and timestamps. The discovery step before search or sync work.
- Delete Index Permanently deletes an index with all records, settings, synonyms, and rules. The most destructive call in this connector, and it never runs without explicit human approval.
- Get Settings Retrieves an index's full configuration: searchable attributes, ranking, faceting, typo tolerance, and more. Captured before every settings change as the rollback record.
- Set Settings Updates index configuration, changing only the provided settings, with optional forwarding to replicas. Paired with Get Settings so changes are reviewable and reversible.
Frequently Asked Questions
What can FlowRunner do with Algolia?
FlowRunner agents can run Search Index, Add Object, and Save Object in Algolia, plus 10 more actions.
Does connecting Algolia to FlowRunner require OAuth?
No. Algolia connects to FlowRunner with an API key, no OAuth flow required.
Can Algolia trigger a FlowRunner workflow automatically?
Algolia doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with Algolia
$100 in credits. No card required. Connect in minutes.