Reporters and researchers often undertake site visits and fact-finding trips to gain the firsthand perspective necessary for telling clear and accurate stories.
What if we could send our own researcher—let’s call him Dr. V—to an ecosystem of AI agents to report its inner workings?
We’ll send Dr. V to visit an augmented call center that uses Teradata’s Enterprise Vector Store. He'll guide us through an extended ecosystem of agentic AI using a real-world example.
The first thing that Dr. V realizes: Even before the call center becomes active for the day, several AI agents have already run in preparation for the day ahead. The company has acquired many leads from external sources (these leads could number anywhere from tens to millions). As soon as the leads persist in the database, the propensity agent runs its predictive AI models to assign a propensity to convert to sold if successfully contacted.
Next, Dr. V observes that the scheduling AI agent runs its predictive AI models to cluster the leads based on the propensity to optimize the call schedule. The reason: A lead may have a high propensity, but the time-of-day or day-of-the-week feature may recommend a call at a specific time or on a specific day.
The call center now has a batch of numbers to be called on optimal days and at optimal times.
Dr. V notices that the identification AI agent builds a segmentation or persona for each lead, based on available features, to customize the best approach in terms of greetings, sales pitch, customer needs, and other criteria.
The call AI agent starts dialing the numbers. When the calls are answered, more features become known, such as whether the contact is using iOS or Android; Samsung or Apple; or a tablet or a smartwatch. The identification AI agent can then rebuild each persona with more information.
Next, each lead on the phone must be scored to determine the next best action to maintain customer engagement and convert the lead. The next-action AI agent constantly recommends the next best response to the potential client on the phone.
As conversations progress and conversions occur, Dr. V sees that the cross-sell AI agent predicts the top two or three additional products to propose to the customer. Cross-pricing AI agents build hyper-personalized optimal pricing for those products.
Should the product or service needed by the customer require a licensed professional (such as a mortgage loan originator), the transfer AI agent has already determined which licensed agent has the highest potential to achieve the desired outcome for each customer. So, the transfer AI agent then recommends transferring to the next available licensed professional to optimize the outcome.
Of course, some of the AI agents mentioned above, such as the next-action AI agent, continue to work with the engagement even after the transfer.
Dr. V also witnesses how the call center utilizes other agents, such as:
- The analysis AI agent, which understands customer sentiment, such as complaints
- The resolution AI agent, which provides the best resolution to increase retention and reduce churn
For Dr. V, there are still more AI agents to explore in this ecosystem—all handling crucial tasks and optimizing outcomes. So, for now, this story is to be continued.
If you’re ready to start your own exploration of AI agent ecosystems and how they can drive business results, check out the possibilities of Teradata’s Enterprise Vector Store. Unleash the full potential of AI with enterprise-scale vector management, unlocking billions of vectors and delivering cost-effective, scalable AI and CX innovation with seamless data integration.