1.2 — Decomposing Complex Requests
CCAO-F Domain 1 · Task 1.2
What You Need to Know
Complex work gets Claude’s best output when split into pieces it can do properly, one at a time.
Four Techniques (Lightest to Heaviest)
- Numbered steps inside one prompt — for connected parts that must follow a sequence
- Staged prompting in one conversation — for a deliverable built in layers; inspect each stage before the next
- Plan-first prompting — when the right structure is not obvious; review a plan before drafting
- Separate conversations per sub-task — for genuinely independent pieces
Choosing the Technique
- Connected parts → one conversation, staged
- Independent deliverables → separate chats
- Structure unclear → plan first
Carrying Context Forward
A new conversation does not automatically contain another conversation’s working material. Memory saves context about you as individual topics, not a transcript. Explicit persistence is required: paste the output, re-attach the file, or keep it in a Project’s knowledge base.
Exam Traps
- The do-everything mega-prompt
- Fragmenting connected work across separate conversations
- Plan-first prompting where the structure is already known and agreed
Check your understanding
Practice Scenario
A programme director must produce a 20-page annual impact report: an overview plus five sections drawn from five programme teams' notes, all in one consistent voice. She wants to approve each part as it's produced and never re-explain settled decisions to a new chat. How should the work be structured?
Log in to save your progress.
Sources
- Prompting best practices, Claude Platform Docs
- Building effective AI agents, Anthropic
- AI Fluency: Framework & Foundations, Anthropic Academy