Workflow · Trigger · Owner · Fallback
Design the safe automation
Pick one repeated job and write the contract before tools enter
Choose the workflow, map trigger/input/output/owner, decide human versus app versus AI, write the fallback, add the failure check, choose the first tool, and assemble the brief.
Worth knowing:The first automation target should be the clearest workflow, not the flashiest one.
OperatorsFoundersManagersAdmins
Start Module 1 →Forms · Zapier · Make · Airtable
Build in existing apps
Use the tools you already trust before adding agent behavior
Start with forms, Sheets, email, and folders. Then build with Zapier, Make, and Airtable, test sample records, and read the run history.
Worth knowing:The first live run should be boring because the sample records already proved the path.
Ops teamsFreelancersSmall teamsBusiness owners
Start Module 2 →Prompt packet · Approval · Privacy
Add AI safely
Let Claude help where language judgment matters, with checks
Create prompt packets, add Claude for drafting or classification, put approval before risky actions, write retry rules, protect private data, and decide when Cowork or Routines belong.
Worth knowing:AI belongs in a step with an output check, not in charge of the whole workflow.
Ops leadsClient teamsAssistantsBuilders
Start Module 3 →Dashboard · Review · Handoff · Rollback
Operate and maintain
Keep the automation useful after the first working run
Build the owner dashboard, run weekly review, change the workflow safely, hand it off, watch costs and limits, pause bad runs, and keep a maintenance checklist.
Worth knowing:The work is not finished until someone owns review, limits, and recovery.
OwnersManagersOperatorsTeams
Start Module 4 →