0%
Dive deeper with Ask Ai tool
YouTube's first native conversational AI assistant, integrated directly into the watch page.
the problem
Users frequently have questions or want to dive deeper into a video's content but are forced to interrupt their viewing experience.
This requires them to leave the YouTube watch page to use external search engines or manually scrub through videos to find specific information, creating a disjointed, inefficient, and often frustrating journey.
We created a tool that provides immediate value by answering questions, summarizing content, and offering recommendations.
Ask — YouTube's first native conversational AI assistant, integrated directly into the watch page.
Initially launched as an experiment for Premium users, it allows viewers to ask questions about the video they are watching and receive contextual answers without interrupting playback.
The interface, accessible via an "Ask" button and other entry points, evolved from a simple chat-bubble design to a more polished and integrated panel. To guide users and reduce friction, the solution heavily features suggested and dynamic prompts, which proved to be a primary driver of interaction.
Ask is powered by a succession of Large Language Models, evolving from PaLM2 to more advanced Gemini models (Pro and Flash). The system uses a Retrieval-Augmented Generation (RAG) approach.
how it works
When a user submits a query, a router directs it to the correct backend.
The tool supports multi-turn conversations with text and rich media responses.
Suggested prompts are a key feature, catering to users' preference, for a "lean-back" experience.
When a user submits a query, a router directs it to the correct backend.
The tool supports multi-turn conversations with text and rich media responses.
Suggested prompts are a key feature, catering to users' preference, for a "lean-back" experience.
Impact
Ask feature successfully scaled to 170 million monthly active users and 10 million daily active users.
User satisfaction has been high
Mobile users
The iOS launch was particularly successful, driving twice the engagement time of the Android version.
The project has been influential, establishing foundational UX patterns for conversational AI at YouTube and shaping the company's long-term AI strategy.
