Strong search starts with the right relevance settings, but it's easy for search configuration to move away from best practices over time.

You can now optimise key search settings with AI Assist, in line with your search goals and Algolia’s best practice. To start, ask AI Assist to “improve my search configuration” and select an existing index. Assist will ask a couple of questions on your industry and search goals, which will tailor its recommendations for the following core search settings: Searchable attributes, Custom ranking, Ranking and Sorting, and Typo Tolerance.

Using Algolia’s best practices, your current index configuration and information on your search goals, Assist will analyse your settings to suggest changes. For any of Assist’s suggestions, you’ll see a clear explanation and details of how your settings will change. Users with the Set-Up Search permission can click to apply all or some of Assist’s changes.

If your index is included in any active A/B tests, you can automatically propagate the same optimisations to other indices. These changes can also be shared with replicas from within Assist. All of Assist’s changes will be tracked in the audit log for your visibility.

Public MCP now supports a new multi-query signature with advanced AI search settings. Configure search parameters such as searchable attributes, ranking, and filters independently of index settings, and update them without recreating the MCP. Click here to learn more.

Recommend-based groups let you populate a Smart Group with items from an Algolia's Trending Items Recommend model into a category page or collection. You define the model to query and Algolia injects its top recommendations at the position you specify. We only support the Trending Items Recommend model. Click here to learn more. 

AI Assist now understands your configuration and uses this to provide advice that's tailored and relevant. A question like “Why do I have a high no-results rate?” will be met with targeted guidance on how to change your current set-up. Assist will also provide links to navigate you to relevant parts of the Algolia platform. This allows you to quickly find the relevant solution needed to take action on an Assist conversation. Assist is aware of your application’s current subscription tier and Algolia release version, and will only direct you to features you can access.

Agent Studio is now generally available, providing a managed platform for building, deploying, and operating AI agents grounded in Algolia data. Customers can configure model providers, connect search and recommendation tools, add safety controls, and monitor conversations through a single API and dashboard. This reduces the complexity of building production-ready AI experiences while providing a consistent, scalable foundation for agent development. Click here to learn more.

Agent Studio now generally supports Model Context Protocol (MCP) servers, allowing AI agents to securely connect to external tools and services through a standardized interface. Customers can extend agent capabilities with approved third-party tools while controlling which tools are available and requiring human approval for sensitive actions. This makes it easier to integrate existing systems into AI workflows without building custom integrations. Click here to learn more.

Merchandisers can now create conditionless rules - rules that ‘always apply’ i.e. without a trigger - directly in the Visual Editor. Previously these rules could only be created from the Manual Editor, putting them out of reach for business users. A new Actions menu also lets you create rules unassociated with the current page or query, in the Visual Merchandiser.

DocSearch now has a dedicated DocSearch Hub in the dashboard, bringing DocSearch resources together in one place. Access the Crawler, Ask AI in Agent Studio, indices, API keys, and UI implementation guidance from a single location. This update reduces navigation across multiple dashboard sections and makes it easier to manage and implement DocSearch.

The Visual Editor now supports custom grid layout configurations, letting merchandisers preview results in the same tile pattern used on their storefront. Configure your preferred layout once at the index level patterns like 3-4-3 and see exactly how your products will appear. Reduces guesswork and eliminates the need to validate changes on a separate staging environment.

Dynamic Facets automatically reorders facets based on how shoppers apply filters, helping surface the most relevant filtering options for each query or browsing context. Rankings are updated daily using filter usage data and returned directly in search responses without additional queries. This helps reduce the effort required to find relevant filters, improves filter engagement, and minimises ongoing manual facet merchandising. Click here to learn more.