Design Applications

What this chapter covers

Design Applications is 14% of the Databricks Generative AI Engineer Associate exam. It covers the decisions you make before writing much code: what the application takes in, what it must give back, and which pieces sit in between.

The chapter has two parts: shaping a single model call (prompts that return a fixed format, picking the model task, choosing chain components from the business goal), and designing multi-stage work (which tools an agent needs, in what order, and when an Agent Bricks agent does the job without custom code).

Questions are mostly short scenarios in which the stated requirement (latency, cost, freshness, who may access what, quality signal) decides the answer. They are single-answer and some multiple-answer multiple choice, and short code is shown as text. There are no exhibits.

How the topics fit together

Every design question starts from the business goal. Turn it into a clear input and a clear output, pick the model task that maps one to the other, then add only the components that task needs.

• Prompt questions ask how to get JSON, a label or a fixed layout back every time.
• Task questions ask which model task fits a requirement, such as classification against summarization.
• Component questions ask which chain parts are needed for a given input and output.
• Tool questions ask which tools an agent needs and the order to call them in.
• Agent Bricks questions ask whether Knowledge Assistant, Multiagent Supervisor or Information Extraction fits, or whether custom code is needed.

How to use these pages

Each topic page has the same shape, so you can skim for what you need.

• What it is: the idea in plain words.
• How it works: the mechanism, with tables and examples.
• What to remember for the exam: the facts to memorize and how questions are phrased.
• Common mistakes: the classic traps.
• Check yourself: three questions to test your understanding.