Dataiku
Overview
Dataiku is an integrated enterprise data science and ML platform. From no-code visual workflows to Python/R coding, users of all skill levels collaborate to build, deploy, and operate data projects. Built on top of the LLM Mesh, the platform now centers on covering AI agent generation, debugging, evaluation, and production operation in a single environment.
Key Features
LLM Mesh
A secure gateway for using LLMs in the enterprise. Supports OpenAI, Anthropic, Azure OpenAI, Bedrock, and others, with model routing, PII masking, centralized cost control, and governance.
Cobuild
Describe what you want in plain language and Cobuild generates an end-to-end project — data pipelines, ML models, agents, and applications — delivered as inspectable visual flows.
Agent Development and Evaluation
Dataiku 15, released on August 14, 2026, added real-time visual debugging of agent execution step by step, Agent Skills that can be packaged once and reused across agents, and Agent Review / Agent Evaluation for repeatable measurement and sign-off.
MCP Server and Conversation API
Native MCP server support makes agents discoverable from MCP-compatible clients, while the Conversation API embeds them in custom apps with server-side conversation history.
Semantic Layer Integration
Agents can query unified semantic models built inside Dataiku or imported from Snowflake and Databricks, keeping answers consistent with business context.
Visual ML
Drag-and-drop ML pipelines. AutoML automates model selection and tuning.
Code Support
Python, R, SQL mixed freely with visual workflows.
Enterprise Features
- MLOps — deploy, monitor, retrain
- Data governance — access control, audit, governance policies
- Project management — collaboration, versioning
- Dashboards — deploy as web apps
Broad Data Sources
Snowflake, BigQuery, S3, Azure Data Lake, Databricks, Kafka, etc.
Pricing
Two free options exist: the on-premise Free Edition for up to 3 users without deployment, automation, or governance features, and a 14-day Dataiku Cloud trial for 2 users covering everything except Govern and advanced LLM Mesh. Paid edition pricing is not published and is quoted per customer based on users, feature set, and deployment mode (cloud editions on AWS/Azure/GCP).
Strengths
- ✅ LLM Mesh multi-LLM governance
- ✅ Visual + code hybrid
- ✅ Enterprise MLOps
- ✅ Data governance
- ✅ AutoML for beginners
- ✅ Broad data platform integrations
Weaknesses
- ❌ Too big for small teams
- ❌ Expensive, raising the adoption bar
- ❌ Steep learning curve
- ❌ Limited Japanese docs
Official Resources
- Website: https://www.dataiku.com
- LLM Mesh: https://www.dataiku.com/product/llm-mesh