Your Page

Every box shows an example of the kind of answer it wants. Press the right arrow key in an empty box to take that example and edit it into your own.

Your answers save as you type. Sign in below and they follow you to any device you use. Stay signed out and they live only in this browser, so clearing your browser data removes them.

Copyable prompts and section summaries are in the resource library. For your first run, open The wording for your first comparison.

1. Run Your First AI Agent Without Writing Code

Your practice

Section 1 · How Would You Explain an AI Agent? (Goals, Tools and Choices)Your first explanationDescribe the agent run you watched in your own words. Keep this first version so you can compare it with your explanation at the end.

Section 1 · How Would You Explain an AI Agent? (Goals, Tools and Choices)Compare the explanationsRecord Arun’s explanation below your own. Which actions did you include, and which did you leave out?

Section 1 · Did Your AI Tool Check the Web? (Sources and Search)Your price checkKeep the question, tool, source, and whether the answer matched the source. Record anything you could not establish from the activity shown.

Section 1 · Can You Run Your First AI Agent? (A Research Comparison)Your first comparisonKeep your decision, prompt, result or link, and any unfinished work or question the tool asked.

2. Understand How AI Agents Work (Models, Tools and Memory)

Your practice

Section 2 · Can You Spot an Agent's 4 Moves in Your Work? (Agent Workflows)Look, Think, Do and CheckChoose a regular job that includes a decision. Describe what you look at, decide, do and check.

3. Tell Agentic AI, Chatbots and Workflows Apart

Your practice

Section 3 · How Do AI Agents, Agentic AI, Assistants and Automation Differ?Your agent exampleDescribe an agent you could use in your work. Name its goal, one decision the model might make and one tool action it could take.

4. Check How AI Uses Company Data (RAG, APIs and MCP)

Your practice

Section 4 · How Does AI Answer From Company Files? (RAG and Knowledge Bases)Knowledge base and RAGKeep a plain-language explanation of knowledge base and RAG, and the company passage you would ask the seller to show.

Section 4 · Can You Tell What an AI Product Really Does? (Claims and Permissions)Approval and evidenceUse your agent example from Section 3. Add an action that should wait for your approval and the evidence you would check to confirm it was completed.

5. Direct AI Agent Decisions (Goals, State and System Prompts)

Your practice

Section 5 · What Is the Difference Between a Goal and an Instruction?The purpose of your requestWrite one sentence explaining what you will decide or produce with the answer.

6. Get Better Agent Results With Prompt Engineering

Your practice

Section 6 · Which Tasks Make a Poor First Agent Exercise? (Task Selection)Choose your practice taskName the task, options, information needed, suitability requirements, and a result you can review in the time you have.

Section 6 · Can You Explain an AI Agent Run? (Prompt, Evidence and Judgment)Your request, result and reviewKeep the prompt and output together. Record deciding facts and sources checked, unresolved points, and the correction you would make.

7. Verify AI Results and Set Limits (Hallucinations and Guardrails)

Your practice

Section 7 · Which Agent Actions Need Approval? (Guardrails and Human in the Loop)Actions that need controlList actions in your task that need review or a limit. State the condition for each approval or limit.

Section 7 · Which Tasks Suit AI Agents? (Reliability and Human Checks)Extend your explanationAdd a short paragraph explaining how you check and correct a wrong answer. Keep the action that waits for your approval visible.

8. Compare AI Agent Types (Training, Human Support and Multi-Agent Systems)

Your practice

Section 8 · How Do You Reuse AI Corrections? (Saved Rules and Testing)Your reusable correction ruleName the recurring mistake and the task it affects. Write the rule, keep the original evidence beside it, and record what happened when you tested the original case and one different case.

Test 1: Mixed evidence status

You are preparing a weekly cost report.

Reusable correction rule:
For these weekly reports, show confirmed costs and estimated costs separately. Label every amount as confirmed or estimated, and cite its input label.

Report request:
Apply the reusable correction rule to the inputs below. Use only the information supplied. Do not calculate a combined total.

[Invoice 104]
Amount: USD 100
Evidence status: confirmed

[Delivery estimate]
Amount: USD 20
Evidence status: estimated

Check after the run

USD 100 appears under confirmed costs with the label Invoice 104, while USD 20 appears under estimated costs with the label Delivery estimate. The answer should not combine the amounts.

Test 2: All costs confirmed

You are preparing a weekly cost report.

Reusable correction rule:
For these weekly reports, show confirmed costs and estimated costs separately. Label every amount as confirmed or estimated, and cite its input label.

Report request:
Apply the reusable correction rule to the inputs below. Use only the information supplied. Do not calculate a combined total.

[Invoice 105]
Amount: USD 100
Evidence status: confirmed

[Delivery receipt]
Amount: USD 20
Evidence status: confirmed

Check after the run

both amounts remain under confirmed costs with their input labels. The answer should not invent an estimated cost or combine the amounts.

9. Measure AI Agent Costs and the Work You Still Do

Your practice

Section 9 · What Work Remains After AI Automation? (Preparation and Exceptions)The work that remainsList preparation, review, exceptions and maintenance that people still handle around your chosen task.

Section 9 · How Do You Manage Work Left by AI? (Review Queues and the Last Mile)When the work reaches youAdd when and how each remaining job reaches you, and who handles it.

Section 9 · What Does It Cost to Run an AI Agent? (Tokens and Usage)The cost of your runUse the saved request and result to identify the run. Record plan allocation, account-reported usage, connected-service charges, human time, setup, maintenance, retries and failed runs. Note how often the job occurs and label any unmeasured value.

Section 9 · What Makes an AI Request Reusable? (Prompt Templates)Your reusable requestSave your prompt with a name that describes its job. State the recurring event that starts it and the result needed next. Keep changing inputs visible and record your checking questions beside it.

Section 9 · How Do You Propose an AI Agent at Work? (Business Case and Pilot)Your pilot proposalName the job, trial duration, inputs, quality standard, human actions, exception owner, complete costs and evidence needed to continue, revise or stop.

10. Prepare for AI at Work (Jobs, Skills and Employment Evidence)

Your practice

Section 10 · What Skills Remain When AI Tools Change? (Requirements and Checking)Make the explanation concreteRead your opening explanation. If it remains abstract, add an example from your own work.

11. Grow Your AI Workflows (Client Work and Local Files)

Your practice

Section 11 · How Do You Turn 1 AI Agent Task Into a Workflow?Your next improvementChoose one improvement or repeatable handoff. Keep the original requirement and state the result you will check.

Section 11 · Bonus: What Will You Try Next With AI?Your new explanationRead Your first explanation in Section 1. Then write a new explanation in Section 11 using your task to name a goal, one decision the agent chose and one tool action it performed. Compare the two explanations.