Your First Human + AI Workflow (applied capstone)

Introduction

You've come the whole way: you know what AI is and isn't, how it works, how to team up with it, how to stay safe, and where it fits in everyday life. This module is the one where it all comes together — not as ideas, but as one calm, repeatable workflow you can use on a real task from beginning to end. Nothing new to memorize; you already have the pieces. Here we simply put them in order — five clear steps. By the end of these ~40 minutes you'll be able to say, honestly: I know how to use AI responsibly, from start to finish.

Your workflow — five steps:

1. Choose the task
2. Let AI assist
3. Review it yourself
4. Verify what matters
5. Finalize and own the outcome

Choose the task

After this lesson you will be able to choose a task well: decide whether AI fits, define a clear goal, protect what's private, and recognize when a task must leave the workflow.

A short story. Leila helps run a small tutoring group and wanted to write a warm welcome letter to new families. Before opening any AI tool, she paused for three quick questions: Is this a good task for AI? (Yes — low-risk drafting.) What exactly do I want? (A friendly, clear one-page welcome.) What must stay private? (No family names or contact details.) Thirty seconds of choosing the task well, and everything after went smoothly.

Step 1 — Choose the task is everything that happens before you ask:

• •Decide whether AI is appropriate — is it low-risk, and does AI actually help? Not every task needs AI — used on the wrong task, it can add confusion, errors, bias, or extra work. If it's a consequential decision — health, legal, money, safety, safeguarding (a child or vulnerable person), or one that's hard to reverse — it doesn't belong in this workflow at all; that's for a qualified person.
• •Define the goal — say, in one sentence, what a good result looks like. A clear goal makes a useful result more likely — without one, AI has nothing to aim at.
• •Protect private information — decide what stays out (names, numbers, anything not yours to share) before you type.

Set up before you ask

A helpful analogy. It's like planning a short trip before you leave: check it's worth going, know your destination, and don't pack anything you can't afford to lose. A minute of choosing well saves an hour of wandering.

Your turn. You want help drafting a community newsletter. What's the right first move? (a) Paste everything you have, including members' details, and see what AI produces. (b) Decide if AI fits, name what a good newsletter looks like, and remove anything private first. (c) Ask AI to decide what the newsletter should say. Answer: (b). Choose the task: decide, define, protect — before you ask.

Here's the aha: choosing the task well — a clear goal and a privacy-conscious start — makes a useful result more likely.

A practical example. Leila writes one line — "a warm, one-page welcome for new families, friendly and simple" — and lists what to keep out. Step 1 is done; the ask will be easy.

A common mistake to avoid: rushing straight to the AI tool with a vague goal and private details still attached. A messy start makes a poor result more likely — and can expose information you meant to protect.

What you can do today: before your next AI task, answer three questions out loud — does AI fit? what's my goal? what stays private?

What you can do now: name what Step 1 includes — decide whether AI fits, define the goal, protect privacy (and if it's consequential, it leaves the workflow).

Let AI assist, then review

After this lesson you will be able to let AI assist with a clear ask (Step 2), then review the result yourself (Step 3) — improving it rather than treating the first output as final.

A short story. Owen volunteers at a local library and needed to turn a long set of event notes into a short update. With his task chosen, he asked clearly — "summarize these notes into five short bullet points for volunteers" — and let AI draft it. Then he reviewed it: he read the whole thing, checked it fit his goal and shared nothing private, tightened two lines, and flagged one detail — the event date — as something to verify next.

The working middle of the workflow is two steps:

• •Step 2 — Let AI assist. Ask clearly, in ordinary, specific language (no special phrasing needed), and let AI produce a draft, summary, outline, or options.
• •Step 3 — Review it yourself. Read the whole result. Check tone, usefulness, fit, privacy, and whether it actually answers your goal. Then improve it, edit it, or ask for a revision. Don't treat the first output as final.

Review, don't rubber-stamp

A helpful analogy. It's like cooking with a helper who chops and stirs on request — you taste as you go, adjust the seasoning, and keep what works. The helper is fast; your judgment is what helps make it good.

Your turn. AI gives you a draft. Before you use it, what should you do? (a) Send it straight away — AI produced it. (b) Assume it's wrong and rewrite it entirely yourself. (c) Read the whole thing, check it fits your goal and shares nothing private, and improve it or ask for a revision. Answer: (c). Let AI assist, then review it yourself — the first output isn't the final one.

Here's the aha: letting AI assist may produce a draft quickly; reviewing it — reading, checking fit, improving — is what helps make it something worth keeping.

A practical example. Owen asks clearly, gets a five-bullet draft, reads it fully, tightens two lines, and marks the event date to verify next. Assist from AI; review and improvement from Owen.

A common mistake to avoid: treating the first AI answer as final — using it without reading the whole thing or improving it.

What you can do today: on one small task, let AI assist, then review the full result — improve one thing before you go further.

What you can do now: say Steps 2 and 3 — let AI assist, then review it yourself.

Verify, finalize, and own it

After this lesson you will be able to verify what matters (Step 4), then finalize and own the outcome (Step 5) — and send consequential decisions to qualified people.

A short story. Camila coordinates shifts at a community centre and used AI to draft next week's volunteer schedule. She'd chosen the task, let AI assist, and reviewed it — but she wasn't done. She verified it against everyone's real availability, adjusted one slot for a volunteer's needs, decided it was ready, and sent it under her own name. AI helped; the verification, the decision, and the responsibility were hers.

The last two steps finish the workflow:

• •Step 4 — Verify what matters. Check the important facts, figures, dates, names, claims, and instructions — the parts you'll actually rely on — using relevant people, reliable records, official sources, or other appropriate evidence. This is the Everyday Verification Core. A shown source is evidence to inspect, not proof on its own — and checking improves your confidence, it doesn't guarantee correctness.
• •Step 5 — Finalize and own the outcome. Decide: use it, revise once more, or set it aside. Make the final call, and own what goes out — AI assists; you're accountable. (Capability is not authority; responsibility can't be uploaded.) And the boundary that never moves: a consequential decision — health, legal, money, safety, safeguarding a child or vulnerable person, or one that's hard to reverse — leaves the workflow and goes to a qualified person.

Your decision, your name

A helpful analogy. It's like a chef plating the final dish. Others prepped, but the chef checks it, decides it's ready, and puts their name on it. The kitchen helped; the chef is accountable.

Your turn. You've reviewed and verified your draft. What finishes the workflow — and what if it turns out to be a medical decision? (a) Send whatever AI produced; it's done. (b) Ask AI to make the final decision. (c) You finalize and own what goes out — and if it's a consequential decision like a medical one, it leaves the workflow and goes to a qualified professional. Answer: (c). You decide and own the outcome; consequential decisions go to qualified people.

Here's the aha: the workflow finishes with you — you verify what matters, decide, and own the result; and when a decision is truly consequential, you hand it to a qualified person.

A practical example. Camila verifies the schedule against real availability, adjusts one slot, decides it's ready, and sends it under her name. Verified and finalized by Camila; AI only assisted.

A common mistake to avoid: letting the output become the decision by default — finalizing without verifying, or sending it just because it's there rather than because you decided.

What you can do today: on your next AI task, verify the one thing that matters, then say out loud, "I decide, and I own this."

What you can do now: name Steps 4 and 5 — verify what matters, then finalize and own the outcome.

Bringing it together

• •One workflow, five steps: Choose the task → Let AI assist → Review it yourself → Verify what matters → Finalize and own the outcome.
• •It quietly uses everything you've learned — task-fit, privacy, clear asking, verification, and human judgment. (Modules 1–8)
• •Not every task needs AI, and consequential decisions leave the workflow for qualified people.
• •That's your first complete Human + AI workflow: AI assists; you stay responsible from the first step to the last.

Examples

• •Choose the task: decide AI fits, write your one-line goal, and remove private details — before you ask.
• •Let AI assist: "summarize these notes into five short bullets," not a vague "help with this."
• •Review it yourself: read the whole draft, check it fits your goal and shares nothing private, and improve a line.
• •Verify what matters: confirm the one date or figure you'll rely on against a reliable source.
• •Finalize and own it: decide it's ready, adjust one thing, and send it under your name.

Stories

• •Leila (tutoring-group organizer, Lesson 9.1) — chose the task well before opening any AI tool.
• •Owen (library volunteer, Lesson 9.2) — let AI assist, then reviewed and improved the result.
• •Camila (shift coordinator, Lesson 9.3) — verified what mattered, finalized, and owned the outcome.