Elastic introduces Agent Builder, a toolset that allows developers to build custom AI agents on their own business data in minutes. The solution combines conversational context engineering with built-in tools for relevance and governance.
The reliability of AI agents depends on providing the right context at the right time. With business information spread across documents, emails, applications, and customer feedback, this is a challenge for many organizations.
Agent Builder offers tools that promise to go beyond standard queries via the Model Context Protocol. Users can ask questions in natural language, select indexes, configure searches, and define agent parameters.
The built-in conversational agent lets you interact directly with all Elasticsearch data. Intelligent tooling automatically selects the right index, understands the data structure, and translates natural language into optimized queries, whether semantic, hybrid, or structured.
Customization and secure integration
Developers can build custom tools with the Elasticsearch query language ES|QL. This allows you to determine exactly which data is used for context and gives you detailed control over relevance, accuracy, and security.
You can define fully customized agents with a custom system instruction. You also decide which tools the agent has access to and configure a specific security profile.
External agents and applications can be securely connected via MCP and A2A, while governance is maintained through the Elasticsearch layer. This enables the effective deployment of AI agents without losing control.
Agent Builder is available immediately as a Technical Preview on Elastic Cloud in serverless mode. Version 9.2 will soon bring this functionality to more platforms.
Tip: What exactly is vector search and when should you use it?