Learning outcome
Know when to use prompts, templates, automations, or AI agents, and design safer workflows with approval steps.
Intermediate course
AI Automation and Agents for Beginners
Know when to use prompts, templates, automations, or AI agents, and design safer workflows with approval steps.
Course project
Design one automation blueprint with trigger, inputs, AI step, approval, error handling, and cost guardrails.
Safety rule
Do not paste passwords, payment data, private IDs, customer records, or illegal requests into AI tools.
Course syllabus
Learn, do, quiz, and save proof.
Module 1
Automation basics
1Prompt, template, automation, or agentChoose the right level of automation for the task and risk level.Start labClose lab
1. Learn
This lesson teaches one practical AI habit: Choose the right level of automation for the task and risk level. You will apply it to a real workflow map or automation plan, compare the AI output with the goal, then save a reusable version only after review.
- 1Open Zapier or the closest approved tool for this task family.
- 2Paste the lab prompt, replace the bracketed parts with a real workflow map or automation plan, and answer any clarifying questions.
- 3Review the output for facts, missing context, privacy, risky claims, and whether it solves the original task.
- 4Improve one weak part, then save the prompt, final output, and review note as course evidence.
2. Study the example
Example: use Zapier to complete this task: Classify 10 business tasks by automation type. A strong result names the goal, gives enough context, asks for a specific format, marks assumptions, and includes a human review step before use.
Job seeker use
Use this skill to build safer job-search assets: tailored resumes, LinkedIn summaries, networking messages, company research notes, ethical interview preparation, and application tracking templates.
Student use
Use this skill for study plans, summaries, practice quizzes, class notes, project outlines, and revision checklists without submitting AI work as your own when your school rules prohibit it.
3. Proof to save
Automation decision matrix.
Copy-ready lab prompt
You are helping me complete a practical AI-for-work task. Task: [describe your real task] Goal: Choose the right level of automation for the task and risk level. Tool I may use: Zapier, Make, n8n, OpenAI API, Google Apps Script Audience: [who will read or use the output] Constraints: keep it accurate, private-data safe, and easy to review. First ask up to 3 clarifying questions if needed. Then create the output in a clear structure. End with assumptions, risk checks, and a final checklist before I use the result.
Tools to try
ZapierMaken8nOpenAI APIGoogle Apps Script
4. Quick quiz and checklist
Pick when an agent is too risky.
- Does the prompt include goal, context, audience, and output format?
- Did you remove private, sensitive, payment, legal, medical, or account data?
- Did you check facts, numbers, claims, tone, and missing assumptions?
- Did you save the final prompt, output, and review notes as proof?
Common mistakes
- Asking a one-line prompt with no audience, source, format, or success criteria.
- Pasting private customer, payment, account, legal, medical, or employer-confidential information.
- Using the AI answer without checking facts, numbers, tone, assumptions, and missing steps.
Passing answer key
A passing answer explains which tool you chose, why it fits the task, what context you gave it, how you checked the output, and what you changed before saving the final workflow map or automation plan.
Rubric
- Clear task and audience: 25%
- Useful prompt structure and tool choice: 25%
- Human review, privacy, and safety checks: 25%
- Reusable saved template or work sample: 25%
2Workflow mapsMap trigger, input, action, output, review, and failure path.Start labClose lab
1. Learn
This lesson teaches one practical AI habit: Map trigger, input, action, output, review, and failure path. You will apply it to a real workflow map or automation plan, compare the AI output with the goal, then save a reusable version only after review.
- 1Open Zapier or the closest approved tool for this task family.
- 2Paste the lab prompt, replace the bracketed parts with a real workflow map or automation plan, and answer any clarifying questions.
- 3Review the output for facts, missing context, privacy, risky claims, and whether it solves the original task.
- 4Improve one weak part, then save the prompt, final output, and review note as course evidence.
2. Study the example
Example: use Zapier to complete this task: Draw a workflow for lead follow-up or support triage. A strong result names the goal, gives enough context, asks for a specific format, marks assumptions, and includes a human review step before use.
Job seeker use
Use this skill to build safer job-search assets: tailored resumes, LinkedIn summaries, networking messages, company research notes, ethical interview preparation, and application tracking templates.
Student use
Use this skill for study plans, summaries, practice quizzes, class notes, project outlines, and revision checklists without submitting AI work as your own when your school rules prohibit it.
3. Proof to save
Workflow map.
Copy-ready lab prompt
You are helping me complete a practical AI-for-work task. Task: [describe your real task] Goal: Map trigger, input, action, output, review, and failure path. Tool I may use: Zapier, Make, n8n, OpenAI API, Google Apps Script Audience: [who will read or use the output] Constraints: keep it accurate, private-data safe, and easy to review. First ask up to 3 clarifying questions if needed. Then create the output in a clear structure. End with assumptions, risk checks, and a final checklist before I use the result.
Tools to try
ZapierMaken8nOpenAI APIGoogle Apps Script
4. Quick quiz and checklist
Find the missing approval step.
- Does the prompt include goal, context, audience, and output format?
- Did you remove private, sensitive, payment, legal, medical, or account data?
- Did you check facts, numbers, claims, tone, and missing assumptions?
- Did you save the final prompt, output, and review notes as proof?
Common mistakes
- Asking a one-line prompt with no audience, source, format, or success criteria.
- Pasting private customer, payment, account, legal, medical, or employer-confidential information.
- Using the AI answer without checking facts, numbers, tone, assumptions, and missing steps.
Passing answer key
A passing answer explains which tool you chose, why it fits the task, what context you gave it, how you checked the output, and what you changed before saving the final workflow map or automation plan.
Rubric
- Clear task and audience: 25%
- Useful prompt structure and tool choice: 25%
- Human review, privacy, and safety checks: 25%
- Reusable saved template or work sample: 25%
Module 2
Safe agent design
1Human approval and loggingAdd checkpoints so AI does not send, publish, or delete without review.Start labClose lab
1. Learn
This lesson teaches one practical AI habit: Add checkpoints so AI does not send, publish, or delete without review. You will apply it to a real workflow map or automation plan, compare the AI output with the goal, then save a reusable version only after review.
- 1Open Zapier or the closest approved tool for this task family.
- 2Paste the lab prompt, replace the bracketed parts with a real workflow map or automation plan, and answer any clarifying questions.
- 3Review the output for facts, missing context, privacy, risky claims, and whether it solves the original task.
- 4Improve one weak part, then save the prompt, final output, and review note as course evidence.
2. Study the example
Example: use Zapier to complete this task: Design an approval flow for social content or customer replies. A strong result names the goal, gives enough context, asks for a specific format, marks assumptions, and includes a human review step before use.
Job seeker use
Use this skill to build safer job-search assets: tailored resumes, LinkedIn summaries, networking messages, company research notes, ethical interview preparation, and application tracking templates.
Student use
Use this skill for study plans, summaries, practice quizzes, class notes, project outlines, and revision checklists without submitting AI work as your own when your school rules prohibit it.
3. Proof to save
Approval checklist.
Copy-ready lab prompt
You are helping me complete a practical AI-for-work task. Task: [describe your real task] Goal: Add checkpoints so AI does not send, publish, or delete without review. Tool I may use: Zapier, Make, n8n, OpenAI API, Google Apps Script Audience: [who will read or use the output] Constraints: keep it accurate, private-data safe, and easy to review. First ask up to 3 clarifying questions if needed. Then create the output in a clear structure. End with assumptions, risk checks, and a final checklist before I use the result.
Tools to try
ZapierMaken8nOpenAI APIGoogle Apps Script
4. Quick quiz and checklist
Identify actions that need human approval.
- Does the prompt include goal, context, audience, and output format?
- Did you remove private, sensitive, payment, legal, medical, or account data?
- Did you check facts, numbers, claims, tone, and missing assumptions?
- Did you save the final prompt, output, and review notes as proof?
Common mistakes
- Asking a one-line prompt with no audience, source, format, or success criteria.
- Pasting private customer, payment, account, legal, medical, or employer-confidential information.
- Using the AI answer without checking facts, numbers, tone, assumptions, and missing steps.
Passing answer key
A passing answer explains which tool you chose, why it fits the task, what context you gave it, how you checked the output, and what you changed before saving the final workflow map or automation plan.
Rubric
- Clear task and audience: 25%
- Useful prompt structure and tool choice: 25%
- Human review, privacy, and safety checks: 25%
- Reusable saved template or work sample: 25%
2Cost, privacy, and error controlsReduce unnecessary AI usage, avoid sensitive data leaks, and handle failed outputs.Start labClose lab
1. Learn
This lesson teaches one practical AI habit: Reduce unnecessary AI usage, avoid sensitive data leaks, and handle failed outputs. You will apply it to a real workflow map or automation plan, compare the AI output with the goal, then save a reusable version only after review.
- 1Open Zapier or the closest approved tool for this task family.
- 2Paste the lab prompt, replace the bracketed parts with a real workflow map or automation plan, and answer any clarifying questions.
- 3Review the output for facts, missing context, privacy, risky claims, and whether it solves the original task.
- 4Improve one weak part, then save the prompt, final output, and review note as course evidence.
2. Study the example
Example: use Zapier to complete this task: Add daily limits and retry rules to one automation plan. A strong result names the goal, gives enough context, asks for a specific format, marks assumptions, and includes a human review step before use.
Job seeker use
Use this skill to build safer job-search assets: tailored resumes, LinkedIn summaries, networking messages, company research notes, ethical interview preparation, and application tracking templates.
Student use
Use this skill for study plans, summaries, practice quizzes, class notes, project outlines, and revision checklists without submitting AI work as your own when your school rules prohibit it.
3. Proof to save
Cost-control plan.
Copy-ready lab prompt
You are helping me complete a practical AI-for-work task. Task: [describe your real task] Goal: Reduce unnecessary AI usage, avoid sensitive data leaks, and handle failed outputs. Tool I may use: Zapier, Make, n8n, OpenAI API, Google Apps Script Audience: [who will read or use the output] Constraints: keep it accurate, private-data safe, and easy to review. First ask up to 3 clarifying questions if needed. Then create the output in a clear structure. End with assumptions, risk checks, and a final checklist before I use the result.
Tools to try
ZapierMaken8nOpenAI APIGoogle Apps Script
4. Quick quiz and checklist
Choose the safest fallback for failed AI output.
- Does the prompt include goal, context, audience, and output format?
- Did you remove private, sensitive, payment, legal, medical, or account data?
- Did you check facts, numbers, claims, tone, and missing assumptions?
- Did you save the final prompt, output, and review notes as proof?
Common mistakes
- Asking a one-line prompt with no audience, source, format, or success criteria.
- Pasting private customer, payment, account, legal, medical, or employer-confidential information.
- Using the AI answer without checking facts, numbers, tone, assumptions, and missing steps.
Passing answer key
A passing answer explains which tool you chose, why it fits the task, what context you gave it, how you checked the output, and what you changed before saving the final workflow map or automation plan.
Rubric
- Clear task and audience: 25%
- Useful prompt structure and tool choice: 25%
- Human review, privacy, and safety checks: 25%
- Reusable saved template or work sample: 25%
3Agent readiness reviewReview whether an AI agent idea is ready, too risky, or better as a manual workflow.Start labClose lab
1. Learn
This lesson teaches one practical AI habit: Review whether an AI agent idea is ready, too risky, or better as a manual workflow. You will apply it to a real workflow map or automation plan, compare the AI output with the goal, then save a reusable version only after review.
- 1Open Zapier or the closest approved tool for this task family.
- 2Paste the lab prompt, replace the bracketed parts with a real workflow map or automation plan, and answer any clarifying questions.
- 3Review the output for facts, missing context, privacy, risky claims, and whether it solves the original task.
- 4Improve one weak part, then save the prompt, final output, and review note as course evidence.
2. Study the example
Example: use Zapier to complete this task: Score an agent idea using risk, cost, accuracy, and human-review criteria. A strong result names the goal, gives enough context, asks for a specific format, marks assumptions, and includes a human review step before use.
Job seeker use
Use this skill to build safer job-search assets: tailored resumes, LinkedIn summaries, networking messages, company research notes, ethical interview preparation, and application tracking templates.
Student use
Use this skill for study plans, summaries, practice quizzes, class notes, project outlines, and revision checklists without submitting AI work as your own when your school rules prohibit it.
3. Proof to save
Agent readiness scorecard.
Copy-ready lab prompt
You are helping me complete a practical AI-for-work task. Task: [describe your real task] Goal: Review whether an AI agent idea is ready, too risky, or better as a manual workflow. Tool I may use: Zapier, Make, n8n, OpenAI API, Google Apps Script Audience: [who will read or use the output] Constraints: keep it accurate, private-data safe, and easy to review. First ask up to 3 clarifying questions if needed. Then create the output in a clear structure. End with assumptions, risk checks, and a final checklist before I use the result.
Tools to try
ZapierMaken8nOpenAI APIGoogle Apps Script
4. Quick quiz and checklist
Approve, revise, or reject sample agent plans.
- Does the prompt include goal, context, audience, and output format?
- Did you remove private, sensitive, payment, legal, medical, or account data?
- Did you check facts, numbers, claims, tone, and missing assumptions?
- Did you save the final prompt, output, and review notes as proof?
Common mistakes
- Asking a one-line prompt with no audience, source, format, or success criteria.
- Pasting private customer, payment, account, legal, medical, or employer-confidential information.
- Using the AI answer without checking facts, numbers, tone, assumptions, and missing steps.
Passing answer key
A passing answer explains which tool you chose, why it fits the task, what context you gave it, how you checked the output, and what you changed before saving the final workflow map or automation plan.
Rubric
- Clear task and audience: 25%
- Useful prompt structure and tool choice: 25%
- Human review, privacy, and safety checks: 25%
- Reusable saved template or work sample: 25%
4CRM, sales, and lead follow-up workflowDesign AI-assisted lead follow-up without sending unreviewed or misleading messages.Start labClose lab
1. Learn
This lesson teaches one practical AI habit: Design AI-assisted lead follow-up without sending unreviewed or misleading messages. You will apply it to a real workflow map or automation plan, compare the AI output with the goal, then save a reusable version only after review.
- 1Open Zapier or the closest approved tool for this task family.
- 2Paste the lab prompt, replace the bracketed parts with a real workflow map or automation plan, and answer any clarifying questions.
- 3Review the output for facts, missing context, privacy, risky claims, and whether it solves the original task.
- 4Improve one weak part, then save the prompt, final output, and review note as course evidence.
2. Study the example
Example: use Zapier to complete this task: Map a lead capture, qualification, reply draft, approval, and CRM update workflow. A strong result names the goal, gives enough context, asks for a specific format, marks assumptions, and includes a human review step before use.
Job seeker use
Use this skill to build safer job-search assets: tailored resumes, LinkedIn summaries, networking messages, company research notes, ethical interview preparation, and application tracking templates.
Student use
Use this skill for study plans, summaries, practice quizzes, class notes, project outlines, and revision checklists without submitting AI work as your own when your school rules prohibit it.
3. Proof to save
Sales automation blueprint.
Copy-ready lab prompt
You are helping me complete a practical AI-for-work task. Task: [describe your real task] Goal: Design AI-assisted lead follow-up without sending unreviewed or misleading messages. Tool I may use: Zapier, Make, n8n, OpenAI API, Google Apps Script Audience: [who will read or use the output] Constraints: keep it accurate, private-data safe, and easy to review. First ask up to 3 clarifying questions if needed. Then create the output in a clear structure. End with assumptions, risk checks, and a final checklist before I use the result.
Tools to try
ZapierMaken8nOpenAI APIGoogle Apps Script
4. Quick quiz and checklist
Choose which sales message must be reviewed before sending.
- Does the prompt include goal, context, audience, and output format?
- Did you remove private, sensitive, payment, legal, medical, or account data?
- Did you check facts, numbers, claims, tone, and missing assumptions?
- Did you save the final prompt, output, and review notes as proof?
Common mistakes
- Asking a one-line prompt with no audience, source, format, or success criteria.
- Pasting private customer, payment, account, legal, medical, or employer-confidential information.
- Using the AI answer without checking facts, numbers, tone, assumptions, and missing steps.
Passing answer key
A passing answer explains which tool you chose, why it fits the task, what context you gave it, how you checked the output, and what you changed before saving the final workflow map or automation plan.
Rubric
- Clear task and audience: 25%
- Useful prompt structure and tool choice: 25%
- Human review, privacy, and safety checks: 25%
- Reusable saved template or work sample: 25%
5Support triage and escalation workflowUse AI to classify support requests while preserving human review for refunds, legal, safety, or angry customers.Start labClose lab
1. Learn
This lesson teaches one practical AI habit: Use AI to classify support requests while preserving human review for refunds, legal, safety, or angry customers. You will apply it to a real workflow map or automation plan, compare the AI output with the goal, then save a reusable version only after review.
- 1Open Zapier or the closest approved tool for this task family.
- 2Paste the lab prompt, replace the bracketed parts with a real workflow map or automation plan, and answer any clarifying questions.
- 3Review the output for facts, missing context, privacy, risky claims, and whether it solves the original task.
- 4Improve one weak part, then save the prompt, final output, and review note as course evidence.
2. Study the example
Example: use Zapier to complete this task: Create a support routing map with categories, templates, and escalation rules. A strong result names the goal, gives enough context, asks for a specific format, marks assumptions, and includes a human review step before use.
Job seeker use
Use this skill to build safer job-search assets: tailored resumes, LinkedIn summaries, networking messages, company research notes, ethical interview preparation, and application tracking templates.
Student use
Use this skill for study plans, summaries, practice quizzes, class notes, project outlines, and revision checklists without submitting AI work as your own when your school rules prohibit it.
3. Proof to save
Support triage map.
Copy-ready lab prompt
You are helping me complete a practical AI-for-work task. Task: [describe your real task] Goal: Use AI to classify support requests while preserving human review for refunds, legal, safety, or angry customers. Tool I may use: Zapier, Make, n8n, OpenAI API, Google Apps Script Audience: [who will read or use the output] Constraints: keep it accurate, private-data safe, and easy to review. First ask up to 3 clarifying questions if needed. Then create the output in a clear structure. End with assumptions, risk checks, and a final checklist before I use the result.
Tools to try
ZapierMaken8nOpenAI APIGoogle Apps Script
4. Quick quiz and checklist
Identify which ticket cannot be handled by automation alone.
- Does the prompt include goal, context, audience, and output format?
- Did you remove private, sensitive, payment, legal, medical, or account data?
- Did you check facts, numbers, claims, tone, and missing assumptions?
- Did you save the final prompt, output, and review notes as proof?
Common mistakes
- Asking a one-line prompt with no audience, source, format, or success criteria.
- Pasting private customer, payment, account, legal, medical, or employer-confidential information.
- Using the AI answer without checking facts, numbers, tone, assumptions, and missing steps.
Passing answer key
A passing answer explains which tool you chose, why it fits the task, what context you gave it, how you checked the output, and what you changed before saving the final workflow map or automation plan.
Rubric
- Clear task and audience: 25%
- Useful prompt structure and tool choice: 25%
- Human review, privacy, and safety checks: 25%
- Reusable saved template or work sample: 25%
6Monitoring and improvement loopTrack errors, costs, customer impact, and approval rates after an automation goes live.Start labClose lab
1. Learn
This lesson teaches one practical AI habit: Track errors, costs, customer impact, and approval rates after an automation goes live. You will apply it to a real workflow map or automation plan, compare the AI output with the goal, then save a reusable version only after review.
- 1Open Zapier or the closest approved tool for this task family.
- 2Paste the lab prompt, replace the bracketed parts with a real workflow map or automation plan, and answer any clarifying questions.
- 3Review the output for facts, missing context, privacy, risky claims, and whether it solves the original task.
- 4Improve one weak part, then save the prompt, final output, and review note as course evidence.
2. Study the example
Example: use Zapier to complete this task: Create a weekly automation audit dashboard with stop rules. A strong result names the goal, gives enough context, asks for a specific format, marks assumptions, and includes a human review step before use.
Job seeker use
Use this skill to build safer job-search assets: tailored resumes, LinkedIn summaries, networking messages, company research notes, ethical interview preparation, and application tracking templates.
Student use
Use this skill for study plans, summaries, practice quizzes, class notes, project outlines, and revision checklists without submitting AI work as your own when your school rules prohibit it.
3. Proof to save
Automation monitoring checklist.
Copy-ready lab prompt
You are helping me complete a practical AI-for-work task. Task: [describe your real task] Goal: Track errors, costs, customer impact, and approval rates after an automation goes live. Tool I may use: Zapier, Make, n8n, OpenAI API, Google Apps Script Audience: [who will read or use the output] Constraints: keep it accurate, private-data safe, and easy to review. First ask up to 3 clarifying questions if needed. Then create the output in a clear structure. End with assumptions, risk checks, and a final checklist before I use the result.
Tools to try
ZapierMaken8nOpenAI APIGoogle Apps Script
4. Quick quiz and checklist
Choose which signal should pause the automation.
- Does the prompt include goal, context, audience, and output format?
- Did you remove private, sensitive, payment, legal, medical, or account data?
- Did you check facts, numbers, claims, tone, and missing assumptions?
- Did you save the final prompt, output, and review notes as proof?
Common mistakes
- Asking a one-line prompt with no audience, source, format, or success criteria.
- Pasting private customer, payment, account, legal, medical, or employer-confidential information.
- Using the AI answer without checking facts, numbers, tone, assumptions, and missing steps.
Passing answer key
A passing answer explains which tool you chose, why it fits the task, what context you gave it, how you checked the output, and what you changed before saving the final workflow map or automation plan.
Rubric
- Clear task and audience: 25%
- Useful prompt structure and tool choice: 25%
- Human review, privacy, and safety checks: 25%
- Reusable saved template or work sample: 25%
Certificate evidence
This course contributes to your AI work portfolio.
Finish the lessons, save your prompts and outputs, then use the capstone checklist to show what AI did, what you reviewed, and where human judgment was required.