Netflix Product Sense: Choosing the Job and Scope for an AI Agent
Company: Netflix
Role: Product Manager
Category: Product Design & Strategy
Difficulty: medium
Interview Round: Other
You are a product manager at Netflix. Propose an AI agent for the company and design it: decide who it serves and which job it takes on for them, describe how it works, and explain how you would know it is succeeding.
```hint Make it act, not just talk
Look for a moment where acting on someone's behalf (searching, comparing, coordinating, queuing) removes friction that the existing recommendation rows and search do not, and check that the moment happens often enough to matter.
```
### Clarifying Questions
- Should the agent serve members, or may it serve internal teams or partners, such as the people who manage the catalog?
- Which devices matter most, given that much viewing happens on TVs, where typing is slow and not every remote supports voice?
- Should the agent support newer parts of the business such as the ad-supported plan, live events or games, or focus on the core series and film catalog?
- What may the agent do without asking first: start playback, add to a list, change profile settings, or touch the account and plan?
### What a Strong Answer Covers
- A goal tied to Netflix's business, such as satisfying viewing that supports retention, and a reason an agent beats the current browsing and search experience
- A chosen segment and the specific friction moments it faces, compared against other candidate users
- The agent's capabilities and limits, its interaction model on TV and phone, what it may do on its own, and how it uses catalog and viewing data while respecting privacy
- A prioritized first version and what comes after it
- Success metrics, guardrails and risks, such as wrong or unavailable suggestions, spoilers, inference cost and intrusive prompts
### Follow-up Questions
- How would you show that the agent shortened the time to start watching, rather than adding one more place to browse?
- Two members of the same household want to watch together but have very different taste. How should the agent handle it?
- The agent recommends a title that is not available in the member's country. What went wrong in the design, and how do you prevent it?
Overview: A product sense prompt asking a Netflix product manager to propose and design an AI agent. It tests choosing the users and job the agent serves, explaining why an agent beats existing browsing and search, setting what it may do on its own, and measuring success with metrics and guardrails.