Last quiz of AI Patterns. Eight questions, 80% to pass. Production patterns: typed outputs, custom validation, weighted scoring, planning chains, real-service tools.
When you pass, you have the full Patterns kit. Tool calling, chain-of-thought, chained prompts, validation, self-critique, moderation, evals, agents, multi-tool agents, eval suites, typed outputs, retry-on-bad-output, scoring rubrics, planning chains, agent + Composio integration. Eight tracks down, one to go.
| Lesson | Concept |
|---|---|
| 23 | Typed output via output_type=YourPydanticModel — schema, parse, validate, auto-retry handled by the library |
| 24 | Retry-on-bad-output extended — custom predicates with cross-field validation, specific feedback per failure layer |
| 25 | Scoring rubrics — weighted criteria, numeric score in [0,1] |
| 26 | Planning chain — typed Plan(steps: list[str]), loop-execute, summarise; cap step count |
| 27 | Agent + Composio — wrap toolset.execute_action as an agent tool |
| 28 | Final synthesis — agent + 2 tools + 3-case eval + weighted rubric, final mean score >= 0.7 |
AI Advanced (deferred for v1) adds: embeddings, RAG, model routing, response caching, prompt versioning, production guardrails, fine-tuning. All built on top of these primitives.
Create a free account to get started. Paid plans unlock all tracks.
Last quiz of AI Patterns. Eight questions, 80% to pass. Production patterns: typed outputs, custom validation, weighted scoring, planning chains, real-service tools.
When you pass, you have the full Patterns kit. Tool calling, chain-of-thought, chained prompts, validation, self-critique, moderation, evals, agents, multi-tool agents, eval suites, typed outputs, retry-on-bad-output, scoring rubrics, planning chains, agent + Composio integration. Eight tracks down, one to go.
| Lesson | Concept |
|---|---|
| 23 | Typed output via output_type=YourPydanticModel — schema, parse, validate, auto-retry handled by the library |
| 24 | Retry-on-bad-output extended — custom predicates with cross-field validation, specific feedback per failure layer |
| 25 | Scoring rubrics — weighted criteria, numeric score in [0,1] |
| 26 | Planning chain — typed Plan(steps: list[str]), loop-execute, summarise; cap step count |
| 27 | Agent + Composio — wrap toolset.execute_action as an agent tool |
| 28 | Final synthesis — agent + 2 tools + 3-case eval + weighted rubric, final mean score >= 0.7 |
AI Advanced (deferred for v1) adds: embeddings, RAG, model routing, response caching, prompt versioning, production guardrails, fine-tuning. All built on top of these primitives.