Dataiku

Dataiku is an enterprise AI platform that puts data preparation, machine learning, MLOps, generative AI and agent orchestration behind one governed interface. It is built for organisations that need analysts, data scientists and engineers working in the same project, and it removes the sprawl of stitching notebooks, schedulers, model registries and LLM gateways together separately.

Dataiku, founded in Paris on 14 February 2013 by Florian Douetteau, Clement Stenac, Thomas Cabrol and Marc Batty and now headquartered in New York, sells a single platform that spans the whole enterprise AI lifecycle rather than one stage of it. Work is organised around a visual Flow in which drag-and-drop recipes, SQL, Python and R steps sit side by side, so a business analyst and a data scientist can contribute to the same pipeline; from there the platform covers AutoML, model deployment, monitoring and MLOps. Its generative-AI layer is the LLM Mesh, a governed gateway that abstracts model providers behind one connection layer and adds routing, usage quotas, spend monitoring, response caching, moderation screening and audit trails, so teams can switch or mix models on cost and compliance grounds without rewriting applications. On top of that sit visual and code agents, an Agent Hub for central agent management, plus Cobuild and Reasoning Systems announced through 2026. Dataiku runs self-managed on-premises or in a customer's own cloud, or as Dataiku Cloud on AWS in multi-tenant or single-tenant form with a customer-chosen storage region, and it integrates with AWS, Azure, Google Cloud, Databricks and Snowflake as compute backends. The company reports over 750 customers, among them Roche, Johnson & Johnson, Michelin, Novartis, Aviva and Standard Chartered, roughly $350M in annual recurring revenue, and 1,200-plus staff across thirteen offices. Gartner named it a Leader in the 2026 Magic Quadrant for AI Platforms for Data Science and ML on 22 June 2026, its fifth consecutive year.

Category

Enterprise Platform

Pricing

Subscription, Contact for pricing, Freemium

Target Market

CIOs, CTOs, Chief Data Officers, Data Scientists, Data Engineers, Analytics Leaders

Deployment

Hybrid, Self-hosted, Cloud-first, Multi-cloud

Founded

2013

Headquarters

New York, United States

Team Size

500+

Customers

750+ organisations, including Roche, Johnson & Johnson, Michelin, Novartis, Aviva and Standard Chartered; Dataiku says it is used by 1 in 4 of the Forbes Global 2000 excluding China

• ✓
  Visual Flow

  A drag-and-drop pipeline canvas where visual recipes, SQL, Python and R steps coexist, so analysts and engineers edit one shared project.

• ✓
  LLM Mesh

  A governed gateway abstracting LLM providers behind one connection layer, adding routing, quotas, spend monitoring, caching, moderation screening and audit trails.

• ✓
  AutoML and MLOps

  Automated model training plus deployment, monitoring and retraining, so models reach production and stay observed rather than dying in notebooks.

• ✓
  Agent Hub with visual and code agents

  Build multi-step AI agents visually or in code and manage them centrally, under the same governance the ML side already had.

• ✓
  Compute pushdown

  Delegates heavy processing to Snowflake, Databricks, Spark or the cloud warehouse rather than running large datasets through the DSS engine.

• ✓
  Governance and sign-off

  Model registry, approval workflows, risk scoring and audit evidence for regulated deployments, sold above the base platform tiers.

• •
  Regulated model deployment

  A bank builds credit or fraud models in Dataiku and uses governance workflows for documented sign-off, versioning and audit evidence before production.

• •
  Multi-provider generative AI rollout

  An enterprise routes internal LLM traffic through the LLM Mesh to cap spend, screen prompts and swap providers without rewriting applications.

• •
  Analyst self-service on governed data

  Business analysts prepare and join warehouse data with visual recipes while engineers keep schema, lineage and permissions under central control.

• •
  Demand forecasting and predictive maintenance

  Manufacturers and retailers train forecasting models on warehouse data, schedule scoring pipelines, and monitor drift as demand patterns shift.

• •
  Consolidating scattered notebooks

  Teams migrate ad-hoc Jupyter work into scheduled, versioned Flows with lineage, replacing scripts that nobody else can run or reproduce.

Dataiku publishes no list pricing whatsoever, so every figure in circulation is a third-party estimate: independent analysis puts entry deployments near $4,000 per month and enterprise contracts well into six figures a year. Seats are role-tiered, with Designer licences priced far above reader licences, and support tier, compute, and custom integration work are all billed on top of the platform. A free self-hosted edition and a 14-day, five-user managed cloud trial exist for evaluation, but anything production-grade requires a sales conversation and an annual commitment.

This page was written from 11 sources, 6 on domains other than dataiku.com.

1. 1.dataiku.com — productvendor
2. 2.dataiku.com — trustvendor
3. 3.dataiku.com — llm meshvendor
4. 4.dataiku.com — get startedvendor
5. 5.dataiku.com — dataiku named a 5x leader by gartner in the magic quadrant fvendor
6. 6.doc.dataiku.com — index
7. 7.sacra.com — dataiku
8. 8.en.wikipedia.org — Dataiku
9. 9.mammoth.io — dataiku pricing
10. 10.aws.amazon.com — B017MTTNFO
11. 11.github.com — dataiku