Make's 2026 AI agent rollout adds native RAG and agentic HTTP requests as built-in capabilities

The native RAG allowing documents and URLs to be uploaded directly for agent context without requiring an external vector database is the accessibility change that makes knowledge-grounded agents buildable by Make users who are not data infrastructure specialists. Uploading your company documentation, your product specs or your process guides as agent context changes what the agent can answer and act on.

The agentic HTTP requests allowing agents to call any external API directly without pre-built Make integrations is the extensibility that removes the integration library as a constraint. An agent that can make authenticated HTTP requests to any API endpoint has access to any system that exposes an API, regardless of whether Make has built a native connector for it.

The multi-turn conversations being supported for complex interactions requiring back-and-forth context is the conversational depth that changes the agent from a single-turn task executor to something capable of handling nuanced requests that develop across multiple exchanges.

For developers building on Make: has the native RAG eliminated cases where you were previously building external vector databases just to give Make agents context?