Lesson 10: Using AI Well at Work — Role-Based Examples
Lesson overview
The ideas in this course become most useful when you can see them applied to work that looks like yours. This lesson provides realistic examples across several common roles — not as rigid scripts, but as demonstrations of how the course principles apply in different job contexts.
What this means
The same underlying skills — giving context, naming the audience, setting quality standards, asking for outcomes, using constraints, and iterating — apply across every role. What changes is the specific tasks and the specific priorities each role brings.
This lesson is meant to help you see yourself in the material and identify what "working well with AI" looks like for your actual job.
Product management
Product managers use AI to think, plan, communicate, and align. The typical traps: prompting for a document instead of an outcome, skipping audience on internal communications, and accepting first-draft language that is too generic for real stakeholder use.
Example task: Feature prioritization framing
I am preparing a prioritization memo for our quarterly planning meeting. Audience: our product team of 8 and one VP who will make final call. The question we are trying to answer is: should we build deep integrations this quarter or focus on core feature improvements?
Write a one-page framing document that lays out the two options, their tradeoffs, and the key open questions we need to resolve before deciding. Do not make the decision — just frame it clearly. Tone: direct, structured, no fluff. Format: two columns for tradeoff comparison, followed by 3–5 open questions.
What makes this work: Clear audience, decision context, explicit outcome (frame, not decide), format guidance, and tone instruction.
Marketing
Marketers use AI for copy, campaigns, outlines, messaging, and rewriting. The typical traps: copy without brand voice guidance, writing for a vague "customer" instead of a specific person, and accepting inflated promotional language without constraining it.
Example task: Campaign email
Write a short promotional email for Black Friday. Audience: past customers who bought from us 6–12 months ago and have not returned. Product: online project management software. Offer: 40% off first 3 months of an annual plan.
Tone: direct and friendly — not hyped up. No "amazing deal" language. Write to someone who already knows us and just needs a credible reason to come back. Under 150 words. Subject line should be specific and clear, not clickbait.
What makes this work: Narrowed audience within customers, brand voice guidance, explicit exclusion of hypey language, length and subject line instructions.
Support and documentation
Support and documentation teams use AI to simplify complex information, draft help content, and make technical topics accessible. The typical traps: documentation that assumes too much knowledge, instructions that skip context for the frustrated user, and language that is technically accurate but confusing.
Example task: Help article
Write a help article explaining how to reset a password in our app. Audience: non-technical users who are locked out of their account and are probably frustrated. Assume they have never done this before. Tone: calm, reassuring, and step-by-step. Avoid technical language. Start with the most direct path to getting back in. Under 200 words.
What makes this work: The emotional context of the reader (frustrated, locked out) shapes the tone. The expertise level (non-technical, never done this) shapes the depth. The instruction to start with the most direct path prevents AI from leading with background information.
Operations and project coordination
Operations teams use AI for process documentation, internal communications, status updates, and coordination tasks. The typical traps: communications that are too long, process docs that assume familiarity, and updates that bury the important information.
Example task: Internal status update
Write a brief weekly status update for a cross-functional team of 12. This week's update covers: the migration to the new CRM is on track for Friday, the vendor contract is pending legal review (no action needed from the team), and the onboarding checklist needs two more reviewers by Thursday.
Format: three bullets, one per item. Each bullet: what is happening, what the status is, and if there is an action needed, who needs to do it. Tone: matter-of-fact. Under 100 words total.
What makes this work: The outcome is clear (inform and prompt action where needed). Format is specified. Length is capped. The three-part structure per bullet is explicitly defined so AI does not invent its own structure.
Developers
Developers use AI for planning, explanation, summarization, documentation, and implementation support. The typical traps: implementation-only thinking (asking for code when a plan would be more useful first), skipping explanation context when writing for non-technical readers, and under-specifying the technical constraints.
Example task: Technical explanation for a non-technical audience
Explain to a non-technical project stakeholder why we are moving from REST to GraphQL for our API. Audience: a product manager who understands what APIs do but not how they work. Focus on what this change means for them practically — speeds up development of new features, reduces the chance of over-fetching data, makes it easier to add new data fields without breaking existing clients. Do not explain how GraphQL works technically. Under 200 words.
What makes this work: The technical instinct to explain the mechanism is explicitly overridden in favor of business impact. The audience's existing knowledge level is defined. The outcome (stakeholder understanding without needing to understand the implementation) is clear.
Example task: Sprint planning summary
Summarize the following issue list into a sprint planning summary for a team of four engineers. Highlight which items are carry-overs from last sprint, which are new, and which have dependencies that need to be resolved before work can start. Format: three sections (carry-overs, new items, blocked items). One line per item. Do not include issue IDs.
[Paste issue list]
Leadership and internal communication
Leaders and stakeholders use AI for communication framing, decision memos, board summaries, and alignment documents. The typical traps: writing at the wrong detail level for the audience, using uncertain language in contexts that require confidence, and producing documents that describe work instead of driving decisions.
Example task: Decision memo
Write a one-page decision memo recommending we shift our support team from a generalist model to a tiered specialization model. Audience: the founding team. They need to decide whether to approve this change at the next all-hands.
Frame: what the current model is, what its key limitation is, what the recommended change is, and why now. Close with the specific decision we are asking for. Tone: direct and confident. No hedging. Under 400 words.
What makes this work: Outcome is a decision, not just information. The frame is prescribed. The close (specific ask) is required. Hedging is explicitly excluded.
Practical exercise
Choose the role from this lesson that is closest to your own. Pick one of the example tasks and substitute it with a real task from your current work.
Write the prompt using the same structure as the example: audience, outcome, tone, format, constraints. Submit it and evaluate the output against the criteria we have covered throughout the course.
Then write a second, refined version of the prompt based on what the first output revealed was missing or unclear.
Reflection prompt
- Which of the role examples felt most like your work? What would you change to make it fit better?
- Are there aspects of your role that AI could help with more than it currently does? What is the prompt design missing?
- What is the one lesson from this course that will change how you use AI most immediately?
Key takeaway
The skills in this course are not abstract. Context, audience, tone, constraints, outcomes, iteration — these apply the same way whether you are a marketer, a developer, a product manager, or an operations coordinator. The work looks different. The approach is the same.