Teradata adds governed agents to Tera AI assistant

The additions reflect a shift in how Teradata is positioning Tera. Tera is an agentic AI workspace and natural-language interface that lets users interact with enterprise data and AI agents without writing SQL queries or code. The update aims to make that workspace usable by business analysts, data engineers and database administrators through natural language requests — for example, asking it to analyze information, build data pipelines or manage infrastructure while working within their organization's access rules.

The new pieces are designed to address two separate problems. The Tera Context Engine connects information from databases, catalogs, pipelines and other sources without requiring companies to move their data, bringing together metadata, data lineage, business definitions and access policies so an agent can interpret a request in the organization's context. Teradata said the engine can trace AI outputs back to their sources and apply the same policies as information moves between systems. Tera Harness, meanwhile, handles execution: selecting the tools, models and data needed, tracking progress across multiple steps, and pausing for human approval before sensitive actions. It also applies controls before an action runs, an approach intended to limit unauthorized or destructive operations. Agent Skills packages common data engineering, analysis and data science tasks into reusable functions.

The stake here is whether enterprises can hand agents enough information and authority to finish work without losing control of data access and approvals. Teradata's own framing, via chief product officer Sumeet Arora, is that most enterprises are not starting from scratch but are dealing with tools that do not work together and a skills gap that makes those tools hard to use at scale. Whether the Context Engine and Harness deliver on that depends on how well customers configure the business knowledge Tera needs — which is why Teradata is offering AI Services to help identify use cases and set up Industry Knowledge Models. The cost comparisons, meanwhile, rest on Teradata's own benchmark conditions and may not carry over to every enterprise workload.