Algolia Launches Agent Studio

Algolia, a retrieval platform provider, today unveiled Algolia Agent Studio, which places a retrieval engine (comprising hybrid, vector search and keyword search, enriched with rules and personalization) at the center of the agent's workflow.

Agent Studio can be used to do the following:

• Build customer support copilots grounded in knowledge bases, CRM, and ticketing history applications.
• Deliver in-product assistants that adapt to user roles, entitlements, and data environments.
• Create smart e-commerce shopping assistants for shoppers and back-end agents for merchandisers, which combine inventory, pricing, and personalization into conversational shopping journeys.

Agent Studio equips developers and product teams with the following:

• Keyword and vector-based NeuralSearch with rules and personalization.
• Bring your own large language model. Agent Studio handles retrieval, policy, and runtime while remaining decoupled from model providers.
• Model Context Protocol (MCP) support so agents can orchestrate context and tools consistently across stacks.
• Orchestration of tools to connect APIs and actions, enabling agents to retrieve context, reason, and act within governed workflows.
• Ready-made React components that enable teams to embed agents directly into applications.
• Observability with traces, evaluation harnesses, and A/B testing that provide clear visibility into why agents respond the way they do.
• Real-time indexing, schema flexibility, and audit trails.

"The rise of AI agents introduces something entirely new: a class of users that exist alongside humans. A person might issue one or two queries to complete a task. An agent, on the other hand, can issue many thousands of queries to interpret context, to verify output, to refine results, and to orchestrate actions. This explosion in retrieval demand changes everything. Infrastructure built for human-scale interaction is simply not enough; agents require retrieval that is governed, fast, and able to scale to hundreds and thousands of queries per task, per agent, and per second without breaking," said Bharat Guruprakash, chief product officer of Algolia, in a statement.

"Most agent platforms stop at the demo. Agent Studio starts from a different assumption: Agents are not just another search box. They trigger on events, chain multiple retrieval steps, and require continuity over time. Memory becomes essential; without it, every interaction starts again. By grounding both retrieval and memory in Algolia's retrieval infrastructure, agents become accurate, adaptive, and trustworthy in production, where it matters most," Guruprakash added.