Dubai, UAE — March 17, 2026 — Dataiku, the platform for AI success, has announced support for NVIDIA Nemotron open models through a new open-source explainability framework designed for enterprise AI agents. Introduced through Dataiku’s 575 Lab open source office, the framework—called Kiji Inspector™—aims to provide transparency and governance for AI agents operating in critical enterprise workflows.
As organisations increasingly deploy AI agents across operational systems, enterprises are placing greater emphasis on understanding how these systems arrive at their decisions. This need is particularly important in regulated and high-stakes environments where AI outcomes can influence revenue, compliance, safety and customer trust. Kiji Inspector is designed to address this challenge by offering explainability capabilities specifically tailored for enterprise AI agents.
The framework’s first supported model family is NVIDIA’s Nemotron open models. As enterprises build sovereign AI infrastructure and deploy their own AI systems, the integration of Nemotron models with Kiji Inspector provides greater visibility into how AI-driven decisions are made. The system introduces built-in explainability features aimed at addressing the long-standing “black-box” problem associated with many AI models.
At the core of Kiji Inspector is a Sparse Autoencoder that analyses model behaviour at the moment an AI agent selects a tool or action. The system identifies the signals behind those decisions and translates them into clear explanations that teams can trace, validate and understand without slowing down system performance.
Hannes Hapke, Director of 575 Lab at Dataiku, said enterprises are increasingly embedding AI agents into decisions that influence revenue, compliance, safety and customer trust, yet many organisations still lack structural visibility into how those systems reason. He noted that bringing Kiji Inspector to NVIDIA Nemotron models enables organisations to inspect and refine AI explainability before potential risks emerge.
The announcement builds on broader collaboration between Dataiku and NVIDIA to support production-grade generative and agentic AI systems. NVIDIA’s Nemotron models provide the performance and capabilities required for enterprise AI agents, while Dataiku provides orchestration across enterprise data platforms, applications and AI services within a governed framework.
Amanda Saunders, Director of Generative AI at NVIDIA, said scaling autonomous AI agents across enterprise environments requires transparency and accountability. She noted that open models such as Nemotron allow organisations to gain deeper visibility into system operations, and when combined with Kiji Inspector, organisations can better understand the reasoning behind decisions made by AI agents.
Industry organisations are also emphasising the importance of explainability as AI adoption expands. Sampath Reddy, Global Innovation Manager for Data and AI at SLB, said that in energy operations, AI delivers value when engineers can understand and trust the decisions generated by these systems. Clear governance and validated workflows enable teams to integrate AI agents into operational environments with greater confidence.
By extending Kiji Inspector support to NVIDIA Nemotron models, Dataiku enables enterprises to deploy high-performance open-source AI while maintaining transparency and governance standards typically associated with enterprise systems. As AI agents become more autonomous and integrated into enterprise workflows, explainable reasoning is expected to play a critical role in regulatory readiness and long-term AI adoption.
Kiji Inspector for NVIDIA Nemotron is available now.
About Dataiku
Dataiku provides an enterprise AI orchestration platform designed to help organisations build, deploy and govern artificial intelligence systems. The platform enables teams to design analytics, machine learning models and AI agents within a single environment that emphasises transparency, collaboration and governance. Operating across multi-vendor cloud and data infrastructure environments, Dataiku connects the full enterprise AI stack to support scalable AI deployment. Leading organisations worldwide use Dataiku to operationalise AI and deliver measurable business value.
