Teradata unveils AgentBuilder to aid agentic AI development
Like many of its competitors, the longtime data management and analytics vendor is adding tools, such as an MCP server, to help customers build advanced applications.
Teradata on Tuesday introduced AgentBuilder, a suite of capabilities designed to help customers develop agentic AI tools.
Unlike generative AI chatbots that require human prompts to act, agents are AI applications with reasoning capabilities and contextual awareness that allow them to act autonomously.
For example, agents can be trained using a combination of an enterprise's proprietary data and large language models (LLMs) to constantly search data to surface insights that humans might not otherwise discover. In addition, agents can be programmed to take on certain tasks such as documenting workloads and optimizing supply chain management to free up humans for other work.
Because of their potential to improve businesses operations and help employees make better informed decisions, agents are the dominant trend in AI development.
AgentBuilder, scheduled for private preview in early October with general availability planned for early 2025, aims to simplify agentic AI development. Once available, it will be an important addition for Teradata's customers, according to Michael Ni, an analyst at Constellation Research.
"For existing Teradata customers, AgentBuilder adds a strategic bridge from analytics to action," he said. "It's a pragmatic extension that builds on Teradata's customers' needs and gives them a governed way to extend their deep understanding of enterprise and industry models, and to deploy AI agents building on the customers' own data and infrastructure."
Based in San Diego, Teradata is a longtime data management and analytics vendor that expanded into AI development following OpenAI's November 2022 launch of ChatGPT, an event that spurred many enterprises to increase their investments in building AI tools.
Tech giants such as AWS and Google Cloud, data platform vendors including Databricks and Snowflake, and specialists such as Informatica and Qlik are among many other data management and analytics vendors that also added AI development capabilities in recent years.
New capabilities
Following ChatGPT's launch, chatbots and assistants, such as Microsoft Copilot, were the early trend in AI development. In early 2024, agentic AI emerged as the major focus of both enterprises developing AI tools and data management vendors given that proprietary data trains AI tools to understand individual businesses.
Databricks and Snowflake are among the many data management vendors that provide tools so customers can develop agents.
For existing Teradata customers, AgentBuilder adds a strategic bridge from analytics to action. It's a pragmatic extension that builds on Teradata's customers' needs.
Michael NiAnalyst, Constellation Resaerch
Teradata already enables users to build agents on other platforms using data stored in Teradata through a Model Context Protocol (MCP) server. Once available, AgentBuilder will enable development directly in Teradata, furthering the vendor's evolution from a focus on analytics to agentic AI development, according to Sumeet Arora, Teradata's chief product officer.
"The next step is AgentBuilder, which allows people to build agents right where the data is, where the knowledge is, which is in Teradata," he said. "That allows you to keep your AI and knowledge together."
One driving factor behind AgentBuilder's development was to help customers train AI agents with data stored across different environments, not just the cloud, Arora continued. Customer feedback and keeping pace with the growing connection between data and AI also played roles.
"To deliver AI with ROI, we must have the ability to deliver agents on Teradata directly," Arora said.
At the core of Teradata's agentic AI development suite is its MCP server. MCP is an open framework created by AI developer Anthropic to simplify building agents and provide a standard for how agents interact with data sources, including LLMs.
In addition, AgentBuilder features the following:
Support for open-source development platforms Flowise and CrewAI.
Integration with Teradata's data management and analytics platforms, including governance controls to ensure the trustworthiness of the data used to train agents.
Teradata Agents, which are prebuilt templates for developing agents for specific tasks, such as system monitoring.
Given that AgentBuilder provides a framework for agentic AI development, Teradata customers will appreciatethe suite when it is available, according to Donald Farmer, founder and principal of TreeHive Strategy.
"Customers who see Teradata as a mission-critical component of their analytics stack will welcome the integration in this release," he said. "Teradata has customarily done good work in building secure, manageable products, and if AgentBuilder maintains that reputation, it could be appreciated."
However, the components that comprise AgentBuilder are comparable to what its competitors offer, Farmer continued.
"This feels like a 'me too' release from Teradata," Farmer said. "Every analytics vendor is getting onboard the agent train, mostly with similar capabilities. Apart from direct access to data in Teradata, it's not clear what AgentBuilder brings to the table that would be unique or differentiating. This feels like an out-of-cycle, ad hoc release rather than a strategically thought-through feature set."
Among the individual capabilities of AgentBuilder, the Teradata Agents are perhaps its most important, according to Ni. Meanwhile, the collective feature set will provide Teradata customers with a flexible structure for developing agentic AI tools.
"By supporting open agent builder frameworks like CrewAI … Teradata is providing agent-building flexibility," Ni said. "By exposing trusted data, metrics and domain models atop its data foundation, Teradata aligns agentic AI with governance and hybrid deployment needs."
However, like Farmer, Ni said while the development framework is significant for Teradata's existing users, it is not markedly different from what other vendors offer and won't force competitors to respond.
"Teradata's AgentBuilder is evolutionary more than disruptive," he said. "But that's what makes it relevant to its target buyers. As the market rethinks hybrid deployments, rising inference costs and lifecycle governance, Teradata plays to its strengths of trust, control and data proximity."
Looking ahead
After historically focusing on data management and analytics, Teradata aspires to become an AI and knowledge platform that helps customers easily develop autonomous capabilities, according to Arora.
Toward that end, the vendor plans to support retrieval-augmented generation applications, machine learning-based analytics and agents, he said. In addition, economic efficiency and developing agents that can automate tasks within Teradata's own platform usually performed by data engineers and other users are focal points.
"There is going to be a lot of focus on making the platform easier to use, making it autonomous," Arora said. "If we are not autonomous, we cannot be part of an autonomous value chain."
Ni, meanwhile, suggested that Teradata continue to exercise patience as it puts together the pieces for customers to develop agentic AI tools.
Founded in 1979, the vendor has an established history in data governance and data quality. As a result, Teradata would be wise to focus on providing customers with underlying capabilities to create an infrastructure for agents, according to Ni,
"The advantage for Teradata is that it doesn't need to chase every AI trend," he said. "It can watch what sticks in the market, then selectively build what enterprises actually need. In the near term, expect Teradata to double down on governance, security and observability, and roadmap priorities like agent memory, retraining loops and enterprise connectors."
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.
