Planning a Monolith Extraction with an LLM Architecture Partner

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-ai-coding-LLM-monolith-extraction-planning-creators

1. What makes an LLM particularly effective for initial monolith extraction planning?

   1. It automatically deploys extracted services to production
   2. It eliminates the need for code reviews during migration
   3. It can analyze the complete module dependency graph faster than any single engineer
   4. It guarantees zero downtime during the extraction process

2. During seam analysis, an LLM can successfully identify which of the following?

   1. Circular dependencies and hidden coupling between modules
   2. The exact business revenue impact of each extraction
   3. The precise network latency cost of inter-service calls
   4. Which engineering team owns each service boundary

3. An LLM proposes three different extraction cuts with tradeoffs. What does this demonstrate about AI's strengths?

   1. Determining which cut aligns with company politics
   2. Guaranteeing that all proposed cuts will work in production
   3. Automatically resolving conflicts between engineering teams
   4. Generating multiple solution variants with comparative analysis

4. A developer asks an LLM to recommend which service to extract first based on business priority. Why might this be problematic?

   1. The LLM will always choose the smallest service first
   2. Business priority questions require reading the entire codebase
   3. LLMs cannot understand any business-related concepts
   4. The LLM has no knowledge of organizational strategy or business goals

5. What type of output should an architecture team expect from an LLM performing seam analysis?

   1. A final decision document signed by the AI
   2. A guarantee of zero future technical debt
   3. A proposed extraction plan requiring human validation
   4. An automatically executable deployment pipeline

6. When an LLM clusters modules by shared callers, what is it actually identifying?

   1. The exact salary of each developer who wrote the code
   2. Modules that should be deleted entirely
   3. Services that will never need updates
   4. Groups of modules that tend to be invoked together

7. An LLM identifies a circular dependency between modules A, B, and C. Why is this valuable for extraction planning?

   1. Circular dependencies indicate which modules can be deleted
   2. This finding proves the code has no bugs
   3. The AI will automatically refactor the circular dependency
   4. Circular dependencies create extraction barriers that must be resolved first

8. What does the lesson mean by a 'reversible' migration sequence?

   1. All code must be duplicated before moving
   2. The migration can only move code backwards
   3. Reversibility is not important for extraction planning
   4. Each step can be rolled back if problems emerge

9. Why can't an LLM know which boundaries reflect real team ownership?

   1. Team ownership has no impact on extraction planning
   2. LLMs always correctly identify team boundaries
   3. Team ownership is an organizational concept not visible in code structure
   4. Ownership is visible in variable names

10. What is the primary value of asking an LLM to propose three different cut lines?

    1. AI can only think of three options at a time
    2. Three cuts mean the extraction is guaranteed to succeed
    3. The first proposal is always the best one
    4. It provides multiple perspectives for the team to evaluate

11. A product manager wants to use the LLM's extraction plan as the final architecture document. What does the lesson advise?

    1. The plan should be converted to a legal contract
    2. AI plans are never useful for architecture
    3. The plan should be signed by the most senior engineer
    4. The plan is a starting artifact, not a final decision

12. When evaluating an LLM's proposed first PR to ship, what should reviewers focus on?

    1. Whether it eliminates the need for further human review
    2. Whether the AI personally tested the code in production
    3. Whether it represents a safe, incremental change that validates the approach
    4. Whether it includes automatic billing charges

13. An LLM suggests extracting a payment service first. Why might this recommendation be flawed?

    1. LLMs always suggest the wrong service
    2. Extraction should never begin with any service
    3. The LLM cannot assess business priority or revenue impact of services
    4. Payment services are never good candidates for extraction

14. What distinguishes a 'seam' in monolith extraction from a simple code deletion?

    1. Seams are not relevant to extraction planning
    2. A seam is another term for deleting unused variables
    3. A seam requires deleting the entire codebase
    4. A seam is a boundary between services that can be split while keeping both functional

15. Why is it important for the LLM to provide risk profiles for each proposed cut?

    1. The AI is legally liable for incorrect risk profiles
    2. Risk profiles guarantee the extraction will succeed
    3. It helps the team understand potential failure modes and plan mitigations
    4. Risk profiles are required by programming language compilers