AI Agents in Action

In my last article, I discussed how Teradata helps AI practitioners go from strategy to results with trusted data, Trusted AI, and an open and connected ecosystem that enables and empowers Bring Your Own Large Language Model (BYO-LLM) capabilities. 

With this infrastructure in place, creating value using generative AI has led to the emergence of agentic AI and AI agents in the marketplace. Let's ask ChatGPT to give us a brief description of these terms:

• “Agentic AI refers to artificial intelligence systems that can operate autonomously, make decisions, and take actions to achieve specific goals with minimal human intervention. These systems exhibit characteristics like reasoning, planning, adaptability, and goal-directed behavior.”
• “AI agents are software entities that perceive their environment, process information, and act accordingly to accomplish predefined objectives. They can range from simple rule-based systems to advanced models utilizing machine learning and reinforcement learning.” 

Now, let me put on my AI practitioner hat and approach these concepts from the perspective of someone who faces these challenges every day at work. 

Let’s say you have a predictive model in production that runs every night at 3:00 a.m. The model uses the latest data to build merchandise sales forecasts for the next two weeks and create reports for business decision-makers. This is actually a form of agentic AI. However, this model lacks the autonomy to make decisions, such as triggering warehouse-to-store shipments based on the forecast or adjusting prices to achieve higher overall revenue. 

Similarly, a stored procedure triggered by an indicator in the system that runs a piece of code to create a table, report, or email is an AI agent (rule-based). If a condition is met and a predictive or prescriptive model is run on demand (rather than on a fixed schedule), it’s also an AI agent. 

So, what’s new about these technologies? After all, we’ve had such capabilities for a long time, and the challenge has always been to create value from the output of such models. 

The difference today is the ability to extract insights and value from unstructured data, which will account for over 80% of global data in 2025. 

Agentic AI and AI agents will only reach their full potential by bringing both structured and unstructured data together to create additional business value, driving transformation and differentiation in the marketplace. 

Let me share a personal experience: I once needed to determine best practices for converting mortgage leads into mortgage sales. First, I looked for tools to convert unstructured data (chats, phone conversations, emails, and notes) into a structured format to better predict conversion rates and create training materials for a team of loan originators. We only had structured data in terms of how many calls and emails were received and how long it took to resolve them. As you can probably guess, I didn’t have LLMs at the time and, therefore, couldn’t leverage the vast amount of unstructured data to optimize the call center operations. 

So, what would an AI practitioner like to have in this setting? Teradata’s Enterprise Vector Store is the answer. 

The growth of unstructured data will drive the need for billions of vectors. The release of Teradata’s Enterprise Vector Store means no one can handle unstructured data more cost-effectively at scale than Teradata. This release provides a critical building block for agentic AI and unlocks the potential of generative AI by fully integrating the platform’s structured and unstructured data capabilities. Enterprises can now seamlessly store, search, and retrieve billions of vectors, enabling actionable insights and transformative customer experiences across industries. The Enterprise Vector Store also enables sophisticated retrieval augmented generation (RAG) use cases like augmented call centers. 

By building the Enterprise Vector Store into our open and connected hybrid cloud platform, Teradata enables organizations to deliver trusted, agentic AI with greater freedom and flexibility. The transformative power of harnessing structured and unstructured data accelerates customer experience (CX) innovation and meets the growing demands for faster return on investment (ROI) in today’s data- and AI-driven organizations. 

Previously, organizations needed to rely on one-off vendors or limited in-database solutions to manage unstructured data assets, growing costs, and shrinking ROI, due to costly data movement, duplicative tech, and the inability to support tens of billions of data points. 

The Enterprise Vector Store brings the power of Teradata’s hybrid platform—up to 20x less expensive and 62x faster than competitors with more limited deployment flexibility—to the latest, most innovative predictive and generative AI use cases. 

With access to Teradata’s proven hybrid cloud analytics and data platform and cutting-edge AI/ML capabilities (such as Bring Your Own LLM), the Enterprise Vector Store enables organizations to deliver faster ROI, better customer experiences, and greater data freedom by harnessing the full potential of billions of vectors and redefining vector data management. 

I’ve built and deployed many predictive and prescriptive models in my career, all of which used only structured data. I always wished to make them more impactful, more relevant, and more adaptable by using the much larger domain of unstructured data, but at the time, I couldn’t. So I’m extremely excited for my fellow AI practitioners to have this capability available with trusted, agentic AI. 

Teradata VantageCloud and ClearScape Analytics™ with Enterprise Vector Store and BYO-LLM capabilities provide the most efficient, performant, and cost-effective way to achieve the promise of agentic AI and AI agents.