Cross-cutting¶
Patterns that apply in every field we cover: how to frame a prompt, how to verify AI output, and how to save a workflow so it is repeatable.
Patterns¶
- Prompt → Verify → Store: draft with AI, verify against source, save the winning prompt to your workflow library.
- Evaluate before you trust: run any new AI workflow on a known input first; never let the first output be the production output.
- Attribution and sourcing: any claim about tools or model behavior gets a source link.
Cross-cutting topics¶
The taxonomy defines an explicit cross-cutting dimension (ai.* topics, added in taxonomy v1.1) that applies to every field: prompting, agents, automation, evaluation, safety, adoption, data, documents, reporting, vendors, and security. See the content taxonomy and research/taxonomy/taxonomy-v1.1.json.
Playbook template (with PAR-4)¶
The canonical playbook structure described in the sample playbook is the contract every field playbook follows.
Planned topics¶
- The PartlyGood prompt framework
- Evaluating an AI tool for your workflow
- Building your personal AI workflow library