At IAB’s NewFronts last week, Samsung Ads and Samsung TV Plus presented an AI-powered CTV platform called “Performance TV,” a new integration with Amazon Ads for CTV which includes shopping functionality and new creator-led “live” programming.
Read the release. (March 24)
Samsung Ads’ Rui Chen, who is the company’s Senior Director of AI & ML, answered a selection of follow-up questions from tipsheet via email.
tipsheet: How is Samsung Ads uniquely using large language models (LLMs) to service its ad clients?
Rui Chen, Samsung: Samsung Ads uses a variety of AI and ML models including genAI and LLMs to deliver an effective end-to-end experience.
This has the capability to look at a variety of signals including on- screen understanding and incorporating them into Ads delivery to provide more relevant and meaningful experience. Rather than relying solely on traditional signals, LLMs help interpret what’s happening on-screen and align ads to those moments in a more nuanced, human-like
way.
This powers smarter decisioning across the campaign lifecycle — from identifying the right audience to optimizing delivery in real time — and continuously feeds insights back into the system so performance improves with every activation. The key differentiator is that this intelligence is built directly into the TV environment, combining device-level data, content signals and AI to drive measurable outcomes.
With AI-Powered Contextual Targeting and Audience Collectives, can you clarify the AI — is this ML, GenAI?
We use a combination of models for any ad serving and would use a combination of deep neural networks, transformer architectures, and LLMs as appropriate.
- AI-Powered Contextual Targeting leverages real-time machine learning and AI-driven content understanding (including LVM-based interpretation) to analyze what’s happening on-screen and determine the optimal moment, placement and messaging for an ad.
- Audience Collectives are built using machine learning models trained on Samsung’s proprietary, cross-device data to identify users most likely to engage or convert.
Together, this creates a continuous feedback loop where campaigns are dynamically optimized based on live signals, not static segments, blending predictive ML with GenAI-driven contextual intelligence.
Can a Samsung Ads advertiser use Amazon Ads’ identity graph (announced in December) to target in a ‘live’ environment?
Advertisers can leverage Amazon’s rich shopping, browsing and streaming signals through the Amazon DSP integration to inform targeting and measurement across Samsung TV Plus inventory. While activation happens through Samsung’s platform, the integration allows advertisers to tap into Amazon Marketing Cloud insights and apply those audiences in campaigns, including those running in live environments.
Combined with Samsung’s real-time AI optimization, this enables advertisers to pair deterministic commerce data with live, high-attention moments on TV, bringing together identity, intent and immediacy in a way that hasn’t been possible before in CTV.