AI Personalization Engine

Retrieval-Augmented Generation (RAG) to personalize LLM interactions using Segment profiles

Project Status


Our team has completed research on this topic. Twilio may or may not release products related to this research in the future.

Research overview

We believe that generative AI is going to transform the way companies engage with customers, and at Twilio we are focused on how companies can keep those automated experiences delightful for their end customers. Personalization is core to solving this problem.

86% of consumers say that personalized experiences increase their loyalty to specific brands. As AI subsumes more and more customer engagement workloads, the companies that embed personalization into their customer engagements will beat out those who don't.

With the AI Personalization Engine project, we set out to understand how to personalize AI driven customer interactions leveraging relevant content from the Segment customer profile, essentially RAG (Retrieval-Augmented Generation) for customer engagement.

By building this on top of the Segment profile, companies can create automated experiences that pull in customer traits as well as live event data to drive the highest level of situational awareness. With event data, an AI Assistant would be able to immediately service the customer with relevant information without needing to ask for additional context. For example, an AI assistant focused on sales could more seamlessly manage a customer's inquiry if they can reference the customer's most recent browsing history; an AI assistant focused on support could better handle a customer escalation if they have the event data and can track issues like a fraud flag or checkout issue.

Additionally, the security and privacy controls from Segment would extend to the AI driven interaction so you don't have to give unfettered access to customer data to the LLM, reducing the risk of data exposure and increasing the accuracy of response.

Illustration showing a conversation between a support agent and a customer turned into trais for a profile

Learn more

We continue to explore these concepts within the AI Assistants project, and would love your feedback on our approach!

Join the AI Assistants waitlist(link takes you to an external page)
Learn more about AI Assistants