Free 30-minute lesson

Build your first AI workflow for real work.

Take one repetitive task from your week and turn it into a workflow that AI can help run reliably. No coding required. No list of tools to memorize.

30 minutesBeginner friendlyOne reusable canvas

Before you start

A workflow is more than a prompt.

A prompt asks for an output once. A workflow defines when the work starts, what information enters, which decisions happen, what good looks like, and how the next run improves. That is what makes AI useful beyond the demo.

The 30-minute build

Five steps. One useful system.

Pick a task worth systemizing

5 min

Choose something you do at least twice a week, starts with recognizable inputs, and ends in a clear deliverable. Avoid your most complex workflow for the first attempt.

Do this: Write one sentence: Every time ___ happens, I need to produce ___.

Map input, decisions, and output

7 min

List what the workflow receives, the decisions made along the way, and what a useful finished result looks like. This is the system before AI enters it.

Do this: Mark each decision as rule-based, judgment-based, or human-only.

Give AI a narrow job

7 min

Assign AI the steps where context, synthesis, drafting, or classification helps. Keep accountability, sensitive judgment, and final approval with a person.

Do this: Define what AI may decide and what it must escalate.

Build the first workflow prompt

8 min

Give the model an outcome, inputs, process, constraints, output format, and a quality check. Good prompts describe the operating system, not just the request.

Do this: Use the prompt template below with one real example from your work.

Run, inspect, and improve

3 min

Do not ask whether the output sounds impressive. Ask whether it is usable, accurate, and faster to verify than doing the work manually.

Do this: Save one good example and one correction for the next run.

Your first workflow prompt

Describe the operating system.

Replace the bracketed text. Keep the structure for future workflows.

You are helping me run a repeatable work process.

OUTCOME
Produce: [the useful final result]

INPUTS
You will receive: [documents, notes, data, messages, context]

PROCESS
1. [first reasoning or transformation step]
2. [second step]
3. [third step]

DECISION BOUNDARIES
You may decide: [low-risk decisions AI can make]
Ask me when: [unclear, sensitive, or high-impact decisions]

OUTPUT FORMAT
Return: [exact structure, length, fields, or format]

QUALITY CHECK
Before finishing, verify: [accuracy, completeness, tone, evidence]

If required information is missing, do not invent it. List what is missing and ask the smallest number of questions needed.

Download the editable workflow canvas

Worked example

Meeting notes → decisions and action items

Input

Transcript, agenda, project context

AI work

Extract decisions, owners, deadlines, and unresolved questions

Human check

Confirm commitments and anything ambiguous

Output

Decision log, action list, and follow-up draft

You are done when

The output is easier to verify than to recreate manually.

Run the workflow three times before adding tools or automation. Each correction becomes a rule, example, or quality check. That feedback loop is the beginning of a real AI system.

Explore the full course tracks