Curious about AI’s potential, I decided to test it with a personal project. I wanted daily climate news without the hassle of checking multiple websites, so I embark on a journey to create an automated newsletter using Zapier that delivers curated stories directly to my inbox.
Step 1:
I started by building an RSS feed integration in Zapier that would automatically pull climate-related news from various sources. I connected this feed directly to a Google Sheets spreadsheet, which would serve as my data repository. This gave me a centralised place where all incoming articles would be stored with their titles, descriptions, sources, and publication dates.
Step 2:
Next, I needed to transform this raw data into a readable daily email. I created a rule-based Zapier workflow that would scan my spreadsheet daily for new RSS entries and automatically trigger an email through Gmail.
My first attempt was a disaster. The automated emails contained all the RSS data dumped as a massive wall of jumbled text. Headlines, descriptions, and metadata were all mashed together in an unreadable format. It looked more like a data export than a newsletter anyone would want to read.
Step 3:
This is where I brought in ChatGPT to solve the formatting problem. I connected my ChatGPT account to the RSS feed data and created a detailed prompt template. The AI now processes each story and outputs a clean format with:
- Bold, underlined titles
- Concise 2-3 sentence summaries
- Critical analysis of the author’s perspective
- Properly formatted source links
- Publication dates
What started as a jumbled mess of RSS data now arrives in my inbox as a professionally formatted newsletter that actually looks like something I’d want to read.
While improved, it was still unreadable. The formatting was clunky, stories ran together, and it looked nothing like a professional newsletter.
This is where the real magic happened. Instead of accepting “good enough,” I dove deeper into prompt engineering. I got specific about exactly what I wanted:
- Clean line breaks between stories
- Proper HTML formatting that Gmail could actually render
- Bold, underlined headlines that stood out
- A clear introduction highlighting the top story
- Professional newsletter tone throughout
Through multiple iterations and testing, I refined my ChatGPT prompt to be incredibly detailed about formatting, structure, and output requirements. The key was being explicit about every element, this included, HTML tags, spacing and tone.
The AI formatting helped, but markdown wasn’t rendering properly in Gmail. I added a Zapier rule to strip out the problematic formatting and bold subjects. After extensive refinement, I finally achieved the clean result I wanted.
3 Main Take-aways:
1) This project demonstrated how AI tools have democratized automation for non-technical users. What would have previously required coding skills and weeks of development, I accomplished in just a few hours using Zapier and ChatGPT. The barrier between having an idea and actually implementing it has dramatically lowered, making custom solutions accessible to anyone willing to experiment.
2) The biggest lesson was that successful AI automation requires precise, detailed prompting. Simply asking for “formatted newsletter content” produced mediocre results, but being explicit about every element, transformed the output quality. The difference between generic prompts and carefully crafted instructions was night and day.
3) Raw automation is just the starting point. Through relentless iteration and testing, I discovered that polishing the details makes all the difference. The extra effort to perfect formatting transformed a clunky data dump into a newsletter I genuinely look forward to reading each morning.