NEW YORK, NY – APRIL 24, 2025: AI platform firm Dataiku in the present day introduced the launch of AI Brokers with Dataiku with capabilities designed to create and management AI brokers at scale, delivering a brand new class of AI functions powered by analytics, predictive fashions, and brokers, the corporate mentioned.
Dataiku mentioned greater than 20 % of consumers now use Dataiku to combine GenAI into their enterprise and information workflows, with a number of tasks per buyer exceeding 1,000 energetic use circumstances.
As firms proceed their rush to deploy brokers, many create clumsy architectures the place brokers function with out correct IT management, range in high quality and relevance, and are sprawled ungoverned throughout groups. Dataiku solves this by augmenting The Common AI Platform with brokers as true enterprise methods — grounded in trusted information, embedded in operational workflows, linked to all AI inputs, and ruled with the identical rigor as any business-critical asset.
“AI is uncooked energy — and it’s time for firms to take management. Corporations are on the verge of repurposing 20 years of enterprise functions constructed on methods like Snowflake, Workday, and SAP with a brand new layer of AI-native functions. These functions demand a mixture of analytics, fashions, and brokers that solely Dataiku can ship,” defined Florian Douetteau, co-founder and CEO of Dataiku.
“As firms mature from exploring to operationalizing brokers, they’ll rapidly understand the necessity to centralize their creation for correct governance as a substitute of letting them sprawl, constantly optimize their efficiency as a substitute of letting them stale, and totally orchestrate them via present IT belongings as a substitute of elevating technical debt. We’re constructing the required capabilities now in The Common AI Platform for this rising market want,” Douetteau added.
Corporations are already starting to really feel the ache of agentic initiatives all over the place within the group with out rationalization or management. Dataiku helps the central creation of brokers with visible agent, a no-code possibility superb for non-technical enterprise customers, and code agent, a full-code possibility superb for builders — each out there in the identical platform, with key capabilities to make sure governance:
- Managed Agent Instruments for guaranteeing high quality and validation of instruments utilized by brokers.
- GenAI Registry for strategic oversight, with agentic use circumstances centrally reviewed, enterprise worth assessed, and dangers evaluated.
- Signal-offs for danger monitoring, with validation workflows off earlier than shifting to manufacturing.
Agentic ecosystems pose rising safety and governance dangers, particularly as fashions evolve quickly, and prices stay tough to watch. IT groups are caught between constructing and sustaining their very own guardrails, locking right into a single vendor, or permitting open experimentation — every path introducing its personal dangers to scale and management.
Dataiku supplies a strong set of capabilities for IT to orchestrate brokers at scale. These capabilities embrace:
- The Dataiku LLM Mesh to handle mannequin entry throughout all suppliers, both for proprietary fashions (OpenAI, Anthropic, Mistral), via cloud providers (Bedrock, Azure, Gemini), or through open supply fashions like Llama and DeepSeek via self-hosting.
- Dataiku Protected Guard to flexibly outline and apply guardrails.
- Agent Hook up with centralize agent entry throughout the group from a single interface, for both dispatching requests to a single agent or having a number of brokers working in live performance to resolve customers’ requests.
AI brokers are brittle and dynamic methods: they will generate sudden errors, and their efficiency can degrade over time. Corporations adopting brokers want to maneuver to a mindset of steady optimization, the place agent testing and efficiency monitoring is routine. Dataiku supplies a deep set of capabilities for agent observability and efficiency monitoring, together with:
- Hint Explorer for full transparency into agent decision-making, enter/output flows, and debugging.
- High quality Guard to guage and monitor agent efficiency constantly with LLM-as-a-judge, analysis units, and golden datasets.
- Price Guard for real-time utilization monitoring, finances enforcement, and inside rebilling.
With a uniquely agnostic method to agentic AI, Dataiku helps all main cloud environments, mannequin suppliers, and information platforms, together with Snowflake, Databricks, Microsoft, AWS, Google, and extra — permitting brokers to run wherever enterprise information and analytics already dwell. As a result of agent improvement is embedded inside Dataiku’s broader Common AI Platform, clients can totally combine brokers into present information pipelines, MLOps workflows, and mannequin governance processes, eliminating duplication and delivering larger ROI.
