Exercise 1: Upgrade the Morning Brief

Objective: Add an AI step to your existing Morning Brief workflow (from Module 2). The AI step should transform raw weather and news data into a 2-3 sentence plain-English summary that highlights what's worth your attention.

What you're starting with: - A working Morning Brief workflow that fetches weather data and news headlines - An email sent to you each morning with raw data

What you're adding: - An AI module (OpenAI or Anthropic) configured with a system prompt - The AI reads the raw weather/news data and writes a brief summary - Your morning email now contains the AI-written summary instead of raw data

Steps:

1. Get an API key. If you haven't already, sign up for OpenAI or Anthropic. Generate an API key and store it securely.

2. Add the AI module. In your workflow platform (Make, n8n, or similar):

3. Add a new module/node
4. Search for "OpenAI" or "Anthropic"
5. Create a connection using your API key

6. Configure the model (start with GPT-4o-mini or Claude Haiku for cost)

7. Write your system prompt. This defines how the AI should think about the task. Example:

``` You are a morning briefing writer. Your job is to read raw weather and news data and write a brief, practical summary that tells someone what they need to know about the day ahead.

Rules: - Write 2-3 sentences maximum - Focus on actionable details: weather impact (umbrella? jacket?), important news - Be conversational and assume the reader is busy - If weather data is incomplete, do your best with what you have - If all data is missing, write: "No data available today" ```

1. Map the data. Configure the user message to include the raw data from your weather and news steps. Example:

``` Weather data:

News headlines:

Write a brief morning summary. ```

1. Test with hardcoded data. Before connecting real data, test with sample weather and news. Run the AI step 5-10 times. Does the output look good? Adjust the prompt if needed.

2. Connect to real data. Replace the hardcoded test data with mapped fields from your previous workflow steps. Run the complete workflow and verify the email looks good.

3. Document what you learned. Write down:

4. The final system prompt you settled on
5. Any changes you made during iteration (e.g., "initially the AI was rambling, so I added 'maximum 3 sentences'")
6. One paragraph describing the prompt iteration process: how many iterations, what went wrong, how you fixed it

Success criteria: - The workflow runs without errors - The AI-generated summary is 2-3 sentences and mentions actionable weather details - You've iterated at least twice on the prompt based on real output - You have a final system prompt you're happy with