Issue 02 · Paperclip

Build Your First AI Workflow in 30 Minutes

Last issue, you audited your AI stack and found the gap — that one repeated task eating your time. Today, we're automating it.

We'll build the simplest possible version. No fancy tools, no expensive subscriptions. Just a workflow you can run tomorrow.

The Framework: Input → Process → Output

Every AI workflow is three steps:

1. Input: What raw material goes in? (emails, notes, data, recordings)
2. Process: What does the AI do to it? (summarize, extract, format, draft)
3. Output: Where does the result go? (task board, email draft, document, spreadsheet)

Example: The Meeting Notes Workflow

Let's walk through a concrete example:

Input

You just finished a 30-minute meeting. You have a recording (phone voice memo) or rough notes.

Process

Paste your notes into Claude or ChatGPT with this prompt:

Here are my meeting notes. Extract:
1. Key decisions made
2. Action items (with owner if mentioned)
3. Open questions
4. Next meeting topics

Format as a clean summary I can paste into Slack.

Output

Copy the result into your team channel or project board. Total time: 3 minutes instead of 15.

Make It Stick

• Save your prompt: Keep it in a text expander or notes app. Reusable prompts are the foundation of AI workflows.
• Run it 5 times: A workflow isn't real until you've run it 5 times. The first time is an experiment. By the fifth, it's a habit.
• Measure the delta: How long did the task take before? After? Write it down. This number is your justification for everything.

Next issue: the tool evaluation framework — how to decide what's worth paying for.