About Elastic

Elastic, the Search AI Company, integrates its deep expertise in search technology with artificial intelligence to help everyone transform all of their data into answers, actions, and outcomes. Elastic's Search AI Platform — the foundation for its search, observability, and security solutions — is used by thousands of companies, including more than 50% of the Fortune 500. Learn more at elastic.co.

Build Gemini Agents capable of working with complex enterprise data

Making AI agents work with real-world, unstructured data can be challenging. Agents can interact with data, but are often inefficient, costly, and unreliable. Elastic Agent Builder provides the capabilities developers need to make their agents more effective:

  • Contextual retrieval across any enterprise data - MCP tools exposing hybrid semantic, keyword, and vector search over any data, structured or unstructured, with hosted models for embeddings, reranking, and LLMs so your agent always gets the most relevant context.
  • Leverage fast, scalable Elastic index as a context layer to store memory and insights, not just raw data - Write agent outputs, summaries, and enriched facts back into Elasticsearch so your agent builds on what it already knows, turning raw signals into retrievable intelligence over time.
  • Custom tools from your data using ES|QL - Define callable tools that wrap ES|QL queries and expose them over MCP, letting your agent search, filter, aggregate, and compute over your data as needed without custom code.
  • Workflow tools that reach across systems - Define tools that retrieve data and take action. Elastic Workflows can call APIs, write to systems of record, and orchestrate multi-step operations so your agent can take real actions. 
  • Workflows that call subagents - Orchestrate specialized subagents as steps within a larger workflow, each powered by its own dynamically loaded Skills, so you can manage context and cost.

How to Get Started

  1. Sign up for Elastic Cloud Serverless: Get a free Elastic Cloud trial at cloud.elastic.co. Create a Serverless Elasticsearch project — infrastructure and scaling are fully managed, so you focus on your agent, not your cluster. Choose your preferred Google Cloud region.
    You can also access Elastic directly through the Google Cloud Marketplace.
  2. Enable Agent Builder: In your Elasticsearch Serverless project, enable Agent Builder from the Kibana UI. Full setup guide: Get started with Elastic Agent Builder.
    Agent Builder ships with built-in search tools for agentic retrieval and a built-in MCP server — no extra configuration required to get your first tools exposed.
  3. Connect Google Cloud Agent Builder via MCP: Point Google Cloud Agent Builder at the Elastic MCP server endpoint found in the Agent Builder Tools UI in Kibana. Authenticate using an Elasticsearch API key. Your Gemini-powered agent will immediately see all the tools you've defined in Elastic.
    Reference architecture: Implementing an agentic reference architecture with Elastic Agent Builder and MCP
  4. Load and enrich your data: Use Elastic's built-in connectors to pull in data from Google Drive, Confluence, SharePoint, GitHub, databases, and more — or index your own data directly. Elastic's ELSER semantic model runs automatically for hybrid search. As your agent generates insights, write them back into Elasticsearch to build your context layer.
  5. Define your tools: Use Agent Builder's UI to create custom tools backed by ES|QL queries or semantic search. Define Workflows that retrieve data, call external APIs, and invoke subagents. Each tool you define is immediately available to your agent via MCP.
    Elastic Agent Builder tool best practices: Tools documentation
  6. Build, iterate, and submit: Test your agent in the Agent Builder playground or directly in Google Cloud Agent Builder. Submit with a public GitHub repo (open-source license required) and a ~3-minute demo video.

Resources

Documentation

Elasticsearch Labs Blogs

Tutorials and Notebooks

Get Access

 

Connect with Elastic