Strategy question

You need to define Glean’s API and SDK strategy for partners building agentic workflows on top of enterprise data. What product principles would guide where you expose low-level primitives versus opinionated abstractions, and how would you balance openness and extensibility against reliability, governance, and long-term compatibility?

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

Start a mock interview on this question · Mock interview from a job description

What this question tests

Tests defining API and SDK product principles that balance partner flexibility against long-term reliability and governance for an enterprise data platform.

How to approach it

  1. State the core principle: expose low-level primitives where partners need custom logic, and opinionated abstractions where most partners want the same outcome.
  2. Apply it concretely, for example expose raw search and permission-check primitives, but ship an opinionated agent-orchestration SDK for the common agentic-workflow pattern.
  3. Guard governance at the primitive layer specifically, since low-level access to enterprise data is the highest-risk surface, requiring scoped tokens and audit logging by default.
  4. Version primitives conservatively, since partners building on low-level APIs are more exposed to breaking changes than those using higher-level SDKs.
  5. Evolve abstractions faster and more freely, since they can change without breaking partners as long as the interface contract holds.
  6. Set a deprecation policy upfront so long-term compatibility is a stated commitment, not an afterthought once partners depend on it.

What a strong answer includes

Common mistakes

Likely follow-up questions

More strategy questions

More questions from Glean

Learn the skill behind it

Chapters of the AI PM course that teach what this question tests.

Preparing for a specific role?

Book summaries for this kind of question

Browse all 4,000+ questions in the bank