Dataiku Launches Cobuild, an AI Agent to Accelerate AI Project Development and Strengthen Governance

Dataiku, a platform that helps companies develop and manage artificial intelligence (AI), has launched Dataiku Cobuild, an AI agent that helps various teams within an enterprise transform business objectives into production-ready AI projects, without the need to write programming code or bypass the organisation’s applicable governance standards. The launch comes at a crucial moment. Over the past few years, companies have invested in building modern data foundations and designing AI strategies. However, the gap between AI experimentation and operational implementation remains a significant challenge. While code-generating AI has been rapidly adopted to accelerate software development, the results often exceed the capabilities or tools available to most business and governance teams to review them properly. On the other hand, many AI agent creation platforms produce prototypes that are disconnected from enterprise infrastructure. This situation leads to growing backlogs of AI projects, increasing technical complexity, and slower implementation processes. “AI can indeed accelerate the process of developing ideas into deployable solutions. However, a company, especially in the pharmaceutical industry, does not just need technology that looks sophisticated. The solution must be clearly understandable, easily auditable, and safe to use in day-to-day operational activities. Dataiku Cobuild addresses that need,” said Neil Patel, Senior Director, Analytics Experience at Pfizer. Starting from a business need or challenge to be solved, Cobuild automatically translates it into a complete AI project on the Dataiku platform, with governance embedded from the initial stage. Powered by the latest generation of AI models, Cobuild can identify relevant data, construct appropriate workflows, and build the necessary components, from data development plans and machine learning models to AI agents and supporting applications. The results are then displayed in a visual flow that can be reviewed, edited, and approved by all stakeholders before being deployed to the production environment. “AI-driven development is only beneficial if the results can truly be applied in an enterprise environment,” said Clément Stenac, co-founder and CTO of Dataiku. “That means the solution built must also be easily understood by business teams, properly supervised by risk management teams, and ready to be operated by IT teams. Cobuild meets those needs. AI provides the speed, business teams understand the company’s needs, and IT teams retain full control over the implementation.” Cobuild operates entirely within the governance framework and permission mechanisms already available in Dataiku. With this approach, companies can expand the use of AI across various business units without reducing the necessary oversight and control. Furthermore, companies can run Cobuild through Dataiku’s own AI services or connect it to their preferred AI models via Dataiku LLM Mesh. The platform supports various leading AI model providers, including Snowflake Cortex AI, OpenAI, Anthropic, AWS Bedrock, Google Gemini, Microsoft Foundry, and Databricks AI Gateway. This support gives companies the freedom to choose models, manage data storage locations, and maintain compliance with internal policies. Dataiku Cobuild will be generally available to customers starting 18 June 2026.