Pluto TV | AI-Driven Conversational Discovery

Conversational discovery allows viewers to express intent in natural language and receive contextual recommendations.

Focus Areas

Conversational AI | Natural Language Search | Recommendation Logic | AI Explainability | AI Product Strategy

Scope

AI Discovery Prototype | Search & Recommendation Layer | Streaming Catalog Navigation | Emerging Product Capability

Role

Senior Director, Product Design

Organization

Paramount / Pluto TV

Partners

Product | Engineering | Playback | Design Systems | Competitive Intelligence

Product Context

Streaming television interfaces traditionally rely on browsing or keyword search to discover content.

However, research and usage data showed many viewers entered a session without a specific title in mind, often abandoning discovery when they could not quickly find something appealing.

As streaming catalogs expand into tens of thousands of titles, traditional navigation patterns—category browsing, editorial rails, and keyword search—begin to lose effectiveness.

Conversational AI presents a different model: allowing viewers to describe what they want to watch in natural language, shifting discovery from menu navigation toward intent-driven exploration.

This initiative explored how a conversational assistant could help viewers express intent, receive contextual recommendations, and transition smoothly into playback.

Architectural Alignment

The work focused on integrating conversational discovery into the existing platform without disrupting familiar navigation patterns.

Rather than replacing search or browsing, the assistant was introduced within the existing Search entry point, allowing the team to validate behavior while maintaining platform stability.

This approach ensured the feature could evolve alongside established discovery patterns rather than competing with them.

subtle executive light gray abstract background.jpg

Strategic Example

The assistant allowed viewers to express viewing intent conversationally:

Examples included prompts such as:

“Show me something funny.”
“I want a crime series.”
“What should I watch tonight?”

The system translated these inputs into contextual recommendations by combining natural language interpretation with catalog metadata and recommendation logic.

Structured prompts and clarification patterns helped refine intent when requests were ambiguous, allowing the assistant to guide discovery without forcing users to reformulate their queries.

AI Search Lander

Moods Tab

Categories Tab

Product Philosophy & Tradeoffs

Several principles guided the design:

Assist, don’t replace

The assistant augmented existing discovery patterns rather than replacing them.

Build trust through transparency

Recommendation responses were framed clearly so users understood why content was suggested.

Balance flexibility with reliability

Natural language input was supported with structured prompts and clarification flows.

Align with Pluto’s brand voice

The assistant adopted Pluto TV’s emerging “Lovable Emcee” tone, using light conversational prompts such as “Scroll less, watch more. Thank me later.” This ensured the AI felt like a natural extension of the platform rather than a generic chatbot.

Preserve editorial and catalog goals

Recommendations surfaced a broad range of titles rather than focusing solely on popular content.

MVP Scope

The initial release focused on validating conversational discovery behavior.

The assistant supported:

Natural language viewing requests
Guided recommendation responses
Suggestion prompts for common discovery intents
Clarification prompts to refine ambiguous queries

Launching within the Search surface allowed the team to evaluate engagement without disrupting existing navigation.

Organizational Leadership

The initiative required coordination across product, engineering, design systems, and recommendation infrastructure teams.

My role focused on shaping the product vision, guiding early interaction exploration, and aligning teams around how conversational discovery could integrate into the broader streaming navigation ecosystem.

The work established design principles and interaction patterns that informed future AI-driven discovery capabilities.

Outcome

The release validated that viewers were willing to express viewing intent conversationally and engage with guided recommendation flows.

Users who interacted with the assistant were able to transition from intent expression to playback, demonstrating that conversational input could function as a viable discovery pathway.

The system also surfaced a wider range of catalog content, suggesting that conversational intent matching could expose viewers to titles less likely to be discovered through traditional browsing.

The learnings informed the roadmap for expanding AI-driven discovery into additional navigation surfaces and future voice-based interaction.