LinkedIn·Product Manager·Onsite - Product Sense / Strategy·Senior

Senior

Jun 2026

Summary

LinkedIn PM interview, one question that sounds deceptively simple but actually requires you to have thought through this stuff before. No fluff, just a pretty direct strategy question.

Questions Asked (1)

How do you decide whether a problem calls for AI, generative AI, an agent-based system, traditional software, or a simple rules-based approach?

Product StrategyTechnical Trade-offsAdaptability & Ambiguity

Author's notes

I fumbled this a bit at first because I started listing use cases instead of talking about the decision criteria.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Frame your answer around a decision framework that starts with the problem's core requirements—data availability, need for personalization or creativity, tolerance for errors, and scalability—then map those to the simplest technology that can deliver. Emphasize that you always default to the least complex solution that meets user needs, and only escalate to AI or agents when traditional approaches fall short. Use a concrete example from your experience to illustrate the trade-offs.

Pro tip: Show that you understand the cost and maintenance implications of each approach—AI and agents require ongoing data pipelines, monitoring, and iteration, while rules-based systems are cheaper but brittle. Mentioning total cost of ownership and time-to-market will set you apart.

1. Define the problem and success metrics

Clearly articulate the user problem, desired outcome, and how you'll measure success (e.g., accuracy, latency, cost, user satisfaction). This sets the bar for what technology is needed.

2. Assess data and complexity requirements

Determine if the problem requires pattern recognition, natural language understanding, or generation (favoring AI/GenAI), or if it can be solved with deterministic logic and structured data (favoring traditional software or rules).

3. Evaluate error tolerance and adaptability

Consider how costly mistakes are and whether the system must adapt to new inputs over time. High error tolerance and need for adaptability point to AI; low tolerance and stable rules point to rules-based or traditional software.

4. Consider agent-based systems for multi-step autonomy

If the problem involves multiple steps, tool use, or dynamic decision-making with minimal human intervention, an agent-based system may be appropriate—but only if the complexity is justified by the value.

5. Start simple, iterate with data

Begin with the simplest viable approach (e.g., rules or traditional software) and only add AI or agents if it fails to meet metrics. Use A/B testing and user feedback to guide escalation.

Key Points to Mention

Data availability and quality: AI and GenAI require large, high-quality datasets; rules-based systems need well-defined logic.
Cost and maintenance: AI models incur training, inference, and monitoring costs; rules are cheaper but less flexible.
User experience and personalization: GenAI excels at creative, personalized content; traditional software is better for consistent, predictable interactions.
Explainability and compliance: Rules-based and traditional systems are more transparent; AI can be a black box, which may be problematic in regulated industries.
Scalability and latency: Traditional software scales easily; AI may have latency and cost issues at scale.
LinkedIn context: Consider how each approach aligns with LinkedIn's professional ecosystem, such as feed ranking (AI), connection suggestions (traditional ML), or automated messaging (rules vs. GenAI).

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.