From UI-Centric Tools to Headless Data Management for AI Agents
A new wave of enterprise data platforms is shifting from traditional, UI-centric tools to headless data management designed for AI agents. Informatica from Salesforce is positioning its Intelligent Data Management Cloud as a callable services layer that agents can invoke mid-workflow, regardless of whether they run on Google Cloud, Snowflake, or Databricks. Instead of relying on pre-built dashboards, agents can access capabilities such as data quality checks, governance rules, metadata search, and address validation on demand. This approach tackles a persistent source of AI agent data quality issues: duplicate customer records, outdated profiles, and unverified attributes that quietly undermine outputs. By exposing governance and master data services through open protocols like Google’s Agent-to-Agent and the Model Context Protocol, Informatica is effectively turning enterprise data governance into an interoperable fabric. The result is AI agent data quality guardrails that are embedded directly in workflows rather than bolted on after the fact.
Golden Record Publishing, Catalog Tags and Iceberg Governance as AI Controls
Informatica’s latest integrations with Databricks and Snowflake highlight how core governance primitives are being repurposed as AI control mechanisms. Golden Record publishing ensures that agents consume a single, reconciled view of entities such as customers or products, limiting the risk of hallucinations caused by conflicting source systems. On Databricks, Informatica can also extract Unity Catalog tags and expose them inside agent workflows, so policies tied to sensitivity, retention, or domain ownership automatically flow into AI logic. For open table formats, Informatica’s Iceberg governance capabilities give enterprises a consistent way to manage schema evolution, access controls, and data lifecycle while AI agents traverse data lakes. Combined with row-level access policy management propagated from Informatica’s Cloud Data Access Management into Snowflake tables, these features reinforce enterprise data governance at the point of AI decisioning. Together they function as policy-aware context, ensuring agents only act on trusted, compliant data slices.
Dataiku Cobuild and the Rise of Inspectable, Governed AI Workflows
Dataiku’s Cobuild on Snowflake targets a different but related problem: the opacity of AI-generated code. Many AI coding assistants generate workflows whose logic remains buried inside an agent’s reasoning path, making it difficult for business users to review or auditors to trace later. Cobuild converts natural-language business requests into visual workflows for data preparation, machine learning, AI agents, and applications running directly on Snowflake. Each workflow captures lineage, versioning, and approvals as part of the development process, reinforcing governed AI workflows as an architectural requirement rather than a UX nicety. Domain experts, analysts, and technical teams can jointly inspect and refine these flows before deployment, aligning AI agent behavior with existing enterprise data governance policies. Early use cases focus on decision agents that operate on curated Snowflake data, such as inventory risk monitors or fraud detection assistants, where traceability and auditability are as important as speed.
Cloud Data Platforms as Hubs for Trusted Agentic AI
Across these initiatives, a common pattern is emerging: cloud data platforms like Snowflake, Databricks, and Google Cloud are becoming the primary hubs for trusted data delivery to AI applications. Snowflake’s Cortex AI, Databricks Agent Bricks, and Google’s agent ecosystems now sit atop governed data layers provided by partners such as Informatica and Dataiku. Developers building agents in Cortex AI can invoke headless Informatica services directly, while Databricks customers can embed metadata search, address verification, and other microservices via MCP-based servers. On Google Cloud, Informatica’s CLAIRE GPT lets data teams discover assets, evaluate quality, and resolve governance issues through conversational interfaces, effectively turning data governance into a collaborative agent in its own right. As enterprises scale AI agents across functions, they are prioritizing governed AI workflows that maintain lineage, compliance, and access control end-to-end, ensuring agentic AI growth does not come at the expense of trust.