Agent Studio
On this page
Agent Studio is a beta feature according to the Algolia Terms of Service (“Beta Services”).
With the Algolia Agent Studio, you can build generative AI experiences using your Algolia data.
It combines advanced large language models with Algolia Search, making it easier to build safe, responsive, and up-to-date GenAI features in your application.
Quickstart
Follow these steps to get started with the Agent Studio:
- Create an agent in the Algolia dashboard. Go to Generative AI > Agent Studio > Agents and click Create agent.
- Configure your agent:
- Write a prompt to define your agent’s behavior.
- Add tools (for example: Algolia Search) to extend capabilities.
- You can test with the built-in model before adding an API key.
- Add a provider:
Go to Settings, click Add provider, and enter your API key. - Test your agent in the dashboard.
- Publish your agent to make it available for integration.
- Integrate with your app using the API or AI SDK UI.
How Agent Studio works
The Agent Studio brings together agents, tools, and large language models (LLMs) to create a flexible, production-ready generative AI runtime.
Architecture overview
- User Input: The user sends a message or query.
- Agent: The agent interprets the input, selects tools, and orchestrates the workflow.
- Tools: The agent can call tools (for example: Algolia Search) to fetch or process data.
- Provider/Model: The agent uses a connected LLM provider (example: OpenAI, Mistral) to generate responses.
- Response: The agent returns a structured, context-aware answer to the user.
Example flow
- User: asks “What are the best wireless headphones?”
- Agent: Uses Algolia Search tool to query product index.
- Tool: Returns search results.
- Agent: Sends results and context to the LLM.
- Provider (LLM): Generates a natural language answer.
- Agent: Returns the final response to the user.
Key concepts
| Agent | Core logic engine that interprets input, chooses tools, and executes actions. |
| Provider | LLM platform (example: OpenAI, Mistral) you connect with your API key. |
| Model | The specific LLM used for generation. |
| Prompt | Instructions sent to the model. |
| Algolia tools | Extend agent capabilities by calling Algolia APIs. |
| Client-side tools | Build user-facing features, such as rendering results or managing UI components. |
What you can build
The Agent Studio supports a wide range of generative AI experiences, including:
- Shopping assistants that answer product questions and recommend items.
- Content summarizers for extracting key information from large datasets.
- Conversational search interfaces for natural language queries.
- Custom workflows using your own prompts, tools, and integrations.
You can start from a template or build your own agent from scratch.
Bring your own API key
To use the Agent Studio in production, you must provide your own API key for the supported provider. This ensures your usage is secure and isolated.
How to add a provider:
- Go to Generative AI > Agent Studio > Settings.
- Under Providers, click Add provider.
- Choose your provider.
- Choose the region (if applicable).
- Enter your API key.
You can try out the Agent Studio on the Algolia dashboard without setting up an API key. For production use, you must configure your own API key.
Any Algolia requests made through the Agent Studio will be counted toward your total Search usage and may incur additional charges to your account.
Supported providers
The Agent Studio supports OpenAI USA and Europe data residency regions.
In Europe, you must also send requests to the https://eu.api.openai.com/v1 base URL for the request to be processed in the region. This is done automatically if you add your provider from the Algolia Dashboard.
To request support for a specific provider, reach out to the support team.
