Ai Coding And Product Systems easy
In your ChatGPT-style assistant, a user's thread has grown past 100K tokens. Each new reply is slower and more expensive, and the thread is approaching the model's context limit. What should the system do?
- A Summarize older turns into a running summary and keep recent turns verbatim.
- B Keep sending the full history every turn, since the model needs every message to stay coherent.
- C Drop the system prompt and tool definitions to free up room for the full conversation history.
- D Truncate the newest messages first so the original context from the start of the thread survives.
Answer: A. Summarize older turns into a running summary and keep recent turns verbatim.
Past roughly **50K to 100K tokens**, a thread needs **summarization or compaction**: you trade some detail from old turns for bounded cost and latency, while recent turns stay exact. Sending the full history forever (B) feels safest for coherence, but cost and latency grow with every turn and the context limit eventually breaks it. Full history is fine only for short threads that sit well under the budget.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- Your assistant uses a classifier to route queries to a small or large model, which saves 60% of GPU spend. Analysis shows about 20% of hard queries are misrouted to the small model, which then gives confident wrong answers, and regenerate clicks on those threads have doubled. What should you do?
- Your workspace AI retrieves the top-k chunks from a shared vector index and passes them to the model with ACL metadata and a system prompt saying 'only use documents this user can access.' A user just received an answer quoting a private HR page they cannot open. What is the correct fix?
- Your support bot answers billing questions. One user has asked the same refund question three times, and retrieval confidence for that topic is low. What should the bot do next?
- Your AI review system receives a 4,000-line pull request spanning 60 files. Sending the whole diff in one prompt yields shallow, generic comments, and real bugs in files near the middle of the diff are missed. How should you structure the review?
- In your ChatGPT-style assistant, power users keep single threads running for weeks, reaching about 150K tokens. Cost per turn keeps climbing, latency is rising, and the model increasingly misses facts from early in the thread. What should you do?
- Your workspace AI filters retrieval using ACLs stored in the vector index, which is rebuilt nightly. An admin removes an employee from a confidential project page, but for the rest of the day the employee can still get AI answers drawn from it. What is the right fix?