Build the skills, then put them together
These lessons reuse the common curriculum. Open any lesson, and return to the capstone when you are ready to assemble the outputs.
Your saved progress loads in this browser.
Completion records your own checks. It is not a test score or a certification. You can open any lesson.
Start the first lesson- Understand models, assistants, agents and automations
- Define the result before choosing the tool
- Give useful context and trace the claims
- Choose a first task you can judge
- Find the constraints that change the plan
- Choose a platform from the work and constraints
- Write instructions that survive the handoff
- Give every agent a job and an owner
- Write a pilot you can actually evaluate
- Test ordinary work, messy work and when to stop
- Turn an approved source into useful marketing drafts
- Review the output against evidence, not confidence
- Measure the whole job, including review
- Decide whether to adopt, adapt or stop
- Turn the accepted workflow into a playbook
- Make the operating decisions explicit
Your capstone exercise
Build a source-based campaign package
Synthetic capstone: a fictional neighborhood repair workshop needs an informative campaign. The approved sample facts are Saturday, 20 places and tools supplied. The intended audience is neighbors learning a simple repair. Price, location, accessibility details and registration destination have not been confirmed. This is a drafting exercise, not a real campaign or a claim of past results.
Work with an approved public input or a clearly labeled fictional example. Completing the exercise does not authorize access to real systems or publication.
Prepare the inputs
- Use the sample facts above or an appropriately authorized offer description with its current version.
- Name the reader, campaign owner, primary action and approved brand guidance, if available.
- Keep a source list and a separate list of unknown details. Do not insert real customer records into the sample.
- Prepare a baseline observation sheet for source gathering, drafting, review and corrections; leave measurements blank until observed.
Do the work
- Write a working brief naming the audience, useful output, evidence boundary, draft-only actions and completion checks.
- Use the platform workspace to record the existing approved assistant, requirements and unresolved facts. No new service is required solely to complete this exercise.
- Extract supported facts and propose three genuinely different message hypotheses. Choose one with a reason grounded in audience needs rather than invented performance claims.
- Create a short LinkedIn draft and an email draft with distinct reader purposes. Map factual claims to the source and preserve the unknown details as questions.
- Run the complete-source, missing-detail and conflicting-detail evaluation cases. Require a question when the source cannot support a claim.
- Review source fidelity and reader usefulness separately. Record every material correction and the effort required to reach an accepted draft.
- Compare the complete task with the baseline if one exists, then write an adopt, adapt or stop decision. Save the accepted instructions and assign the publishing decision to a real owner.
Check what you made
- One audience and intended action are explicit.
- The three message hypotheses differ in their reason to matter, not merely their adjectives.
- Every material factual statement is supported; no testimonial, result or offer detail is invented.
- Missing price, venue, accessibility or destination details remain unresolved until the source owner confirms them.
- The package contains drafts and review evidence; no sending or publication is implied by finishing the exercise.
- The pilot record includes corrections and review effort, or marks those measurements as not yet observed.
Take these outputs with you
- One-page campaign brief and message hypotheses.
- LinkedIn and email drafts, with a claim-to-source map.
- Evaluation cases, correction log and pilot decision.
- A short reusable playbook naming source, editorial and publishing owners.
Adapt the capstone to your context
Owner-led business or small team
Run researcher, writer and reviewer roles in sequence with one assistant. Read the source yourself before using the draft.
Department or growing team
Assign one writer per output and let an editor review the claim map. Have a colleague repeat the process on a second sample.
Large or complex organization
Use the approved tool, brand and claim sources. Route publication and any restricted claims through existing owners; unresolved approvals remain outside the exercise.
Take this capstone into your plan
Use the exercise as a starting point. Add a real owner, baseline, permissions and review process in My AI Plan.
Run researcher, writer and reviewer roles in sequence with one assistant. Read the source yourself before using the draft.
Open My AI Plan