ThinkAnalytics specialises in delivering personalised content discovery, metadata, viewer insight and targeted advertising for media and entertainment, with its products used by broadcasters and streaming services around the world.
In order to deliver even more personalised recommendations, tailored to individual users, the company worked with AWS, using Amazon Bedrock to add generative AI capabilities to its Think360 content discovery and viewer engagement platform
Peter Docherty, founder and chief technology officer, ThinkAnalytics, talks to TVBEurope about how GenAI helps with personalisation, and why it’s so important in the current streaming market.
How is ThinkAnalytics using generative AI?
Using AI and generative AI is a key priority for ThinkAnalytics as it takes personalisation to the next level. ThinkAnalytics has been using Machine Learning/AI for over 20 years and has always been a fundamental part of our products since day one. Most recently, at IBC 2023, we unveiled voice, chatbot and other solutions that combine generative AI, personalised recommendations, content metadata and natural language interaction.
We are working with broadcasters and content providers to add new ways of engaging with individuals or groups of viewers using generative AI for personalised recommendations and other inbound and outbound interactions including personalised emails and social media.
For example, Think360 uses generative AI to create personalised outbound email and social media messaging to reach users at risk of churn. With generative AI–enriched engagement, Think360 sends those users communications tailored to their interests and preferences. This increases interaction, engagement, and ultimately helps drive retention and reduce customer churn.
How does GenAI help with personalisation?
Think360 personalises the entire viewer experience, from initial content discovery through customer retention and churn prevention. The addition of generative AI means that viewers now have access to personalised recommendations using natural language on a variety of platforms. By using personalised recommendations, personalised emails, and chatbots, providers can keep viewers engaged and reduce churn. They also help reach those dormant users who are at risk of churn.
Using generative AI combined with ThinkAnalytics content metadata and ontology, Think360 adds additional context to the answer. It explains why you might like a movie, for example, because it features one of your favourite actors. Also, it adjusts the tone based on who the end user is. To appeal to younger viewers, for example, the service converses in an animated, enthused tone. GenAI can also be used in the metadata enrichment process to help understand the content at a deeper level, again combined with ThinkAnalytics ontology of over 38,000 keywords/tags, moods, themes, subjects, etc.
Why is personalisation so important in the current streaming market?
Churn is one of the biggest problems facing today’s streamers and pay-TV companies. For example, the first 100 days are critical in turning a new user into a loyal viewer and keep them returning to the service. By offering personalised user experiences, streamers can increase viewer engagement, showing the viewers that there is more great content available to enjoy and ensuring that they return more often and watch a wider variety of content when they are on the platform.
Think360’s AI and ML can also identify signs of churn the second they start onboarding a new user by understanding every nuance of the user journey, including analysing their searches, to detect their preferences and mood to offer compelling content. This is particularly important to give those unregistered users of AVoD and FAST services a reason to return.
How is ThinkAnalytics leveraging technology from AWS as part of this workflow?
ThinkAnalytics uses a wide variety of AWS services, including Amazon Bedrock, a fully managed, serverless platform, to create generative AI–enriched interactions. It also uses the Bedrock’s Amazon Titan and Anthropic Claude models for text-generation capabilities. For the conversational interface, ThinkAnalytics uses Amazon Lex, a serverless AI chatbot that supports text and voice input.
The user interacts by text or voice with the Lex chatbot, which sends the request to an AWS Lambda function. This function then calls the Think360 personalisation engine to generate detailed recommendations and search results. The data is sent to Amazon Bedrock, where a large language model enriches the response and creates a tailored, personalised explanation of why the viewer will enjoy the content.
Why did you choose to work with AWS rather than another provider?
ThinkAnalytics has a long-standing partnership with AWS and has had live customer deployments in AWS for over 10 years. Amazon Bedrock was selected for its variety of techniques available from multiple providers, not only Amazon/AWS, as well as a serverless experience, requiring no capacity management or provisioning of servers to operate. Multiple GenAI techniques are required to implement the wide variety of business solutions that customers require and different GenAI techniques also have different strengths and weaknesses. Using Amazon Bedrock, customers are charged only for what they consume, which makes it well-suited to applications with varying levels of activity.
Do you envisage developing the technology further, and what would that involve?
Yes, this is just the start. Together with AWS we will continue to push the boundaries of what’s possible with AI-driven personalisation, including personalised experiences, improving marketing performance, and boosting customer lifetime value.
