Generative AI & LLMs
AI Agents & Tool Calling
9 practice questions. Free questions open a full answer guide; the rest unlock with Pro.
- When an LLM 'calls a function,' the arguments can be perfectly valid JSON yet semantically wrong — like a plausible but nonexistent order ID. Why does constrained decoding not catch this, and what would you put in place to handle it?
- Your tool-calling agent runs its tools strictly one at a time and feels slow, and a teammate wants to make every tool call run in parallel to speed it up. How do you reason about which calls can safely run in parallel, and which must stay sequential or gated?Go Pro
- When an LLM uses a tool, what actually happens end to end? Walk me through how the model 'calls' a function and gets the result back.Go Pro
- How does an LLM 'agent' actually use external tools or functions, and what would you put in place to keep it reliable when a tool call fails or the agent loops?Go Pro
- When an LLM agent is given several tools, how does it decide which one to call, and what risks does giving it tools introduce?Go Pro
- When would you build an agent that decides which tools to call on its own, versus a fixed pipeline that calls the same steps in a set order? How do you weigh the trade-off?Go Pro
- You're building an LLM agent that calls real tools — search, internal APIs, a database — and it works in demos but is unreliable in production: wrong tool, bad arguments, or it loops. How do you design the agent and its tools to be reliable?Go Pro
- What distinguishes an LLM agent from a single LLM call, and how does the tool-calling loop work?Go Pro
- Your tool-calling agent started with five tools and now needs to choose among fifty, and its tool-selection accuracy is dropping. How do you keep it choosing the right tool at that scale, and how do you measure whether you've actually improved it?Go Pro
Want questions matched to your role? Paste a job title, job description, or CV for a personalized set, or go Pro to unlock the full bank.