Teradata has announced three new capabilities for its Tera platform, Context Engine, Harness and Agent Skills, turning it into what the company describes as an agentic coworker that lets anyone from business analysts to platform engineers analyse data, build AI applications and automate workflows without deep Teradata specialisation.
Tera Context Engine is a vendor-neutral context and orchestration layer that sits above existing data environments, connecting databases, structured and unstructured data platforms, pipeline engines, catalogues, models and AI agents without moving data or displacing existing infrastructure. Its native context graph transforms scattered enterprise data into reusable business knowledge by preserving metadata, lineage, semantics and business meaning as relationships rather than flattening them into relational structures.
Tera Harness is an intelligent execution layer that routes work across the appropriate skills, tools, data and models based on the task at hand. Agent Skills, purpose-built for data engineering, data analysis and data science, provide the execution specialisation that general-purpose AI assistants lack for quantitative enterprise workloads.
Teradata has published benchmark results comparing Tera against baseline approaches. On SWE-bench Pro using the Opus 5 model, Tera consumed 73% fewer tokens than the standard configuration while achieving higher task completion and completing work 42% faster at 58% lower total cost. On data-eng-bench, a benchmark from Snowflake Labs and Bespoke Labs, Tera achieved the highest Pass3 score and delivered 53% lower cost per reliably solved task than Snowflake Cortex Code, based on published benchmark data.
Organisations retain model, data and infrastructure choice; Tera can run across cloud, on-premises and sovereign environments. Teradata is listed on the NYSE as TDC.
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
