datarobot-genai

What it is and what it does

DataRobot GenAI is a framework for building AI agents that integrates with DataRobot's LLM Gateway and deployment infrastructure. It provides a unified entry point (`get_llm()`) that routes requests to DataRobot LLM deployments, external LLM providers, or NVIDIA NIM, backed by LiteLLM. The library supports multiple agentic frameworks—LangGraph, LlamaIndex, CrewAI, and NVIDIA NAT—through a single abstraction layer, eliminating the need to write framework-specific integration code.

Beyond LLM routing, it includes a library of agentic tools (`drtools`), a DataRobot-compatible MCP server (`drmcp`), and orchestration via a low-code `workflow.yaml` interface. Agents expose a standard AG-UI event stream for consistent rendering across the DataRobot platform and custom UIs. The package also provides DRAgent, a serving layer that handles distributed tracing, async generations, and WebSocket communication for real-world deployment.

Use it for:

• Build a LangGraph agent that routes LLM calls through DataRobot's gateway without rewriting framework-specific code.
• Orchestrate multi-agent systems (planner/writer crews) using a shared `workflow.yaml` configuration across different frameworks.
• Deploy an agent with DRAgent to serve it as a web application with distributed tracing and async generation support.
• Integrate agentic tools via `drtools` to give agents first-class capabilities for interacting with external systems.
• Expose a custom MCP server in DataRobot using `drmcp` for standardized tool discovery and invocation.
• Evaluate agent performance using built-in evaluation utilities backed by the NeMo Evaluator launcher.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

A toolkit for building and deploying AI agents on DataRobot, providing unified LLM routing, agentic tools, multi-framework support (LangGraph, LlamaIndex, CrewAI, NIM), and orchestration through a low-code workflow interface.

Yes, if you are building agents on DataRobot or need unified LLM routing across multiple agentic frameworks. The library is actively maintained, has no known vulnerabilities, installs cleanly with low friction, and is permissively licensed. Install one of the framework-specific extras (crewai, langgraph, llamaindex, nat) matching your choice of agent framework. Requires a DataRobot account for production features.

Install

datarobot-genai on PyPI

pip install datarobot-genai

uv add datarobot-genai

poetry add datarobot-genai

Installing datarobot-genai

Before you install

Active maintenance with a release on 2026-08-14. Requires Python 3.11–3.13 and installs as a pure wheel with no runtime dependencies in the base package; framework-specific extras (crewai, langgraph, llamaindex, nat) pull in their respective ecosystems. Install friction is low.

License in practice

Licensed under Apache-2.0 (permissive), allowing commercial use, modification, and distribution with minimal restrictions—suitable for most production and proprietary projects.

Quickstart

pip install "datarobot-genai[langgraph]"

import datarobot_genai
llm = datarobot_genai.get_llm("langgraph")

Requires Python 3.11–3.13. DataRobot-backed features require a DataRobot account and DATAROBOT_API_TOKEN environment variable set.

Verify before relying

• Whether the library's multi-agent patterns (planner/writer crews, LangGraph multi-node graphs, LlamaIndex AgentWorkflow) are production-ready or still experimental.
• Performance characteristics and latency overhead of the unified LLM layer routing through DataRobot LLM Gateway.
• Compatibility and stability of the AG-UI event stream contract across different framework versions.

Package facts

Evidence: datarobot_genai-0.27.14-py3-none-any.whl

Tags

More Artificial Intelligence packages

Further reading