👋 I write practical tips for SMEs on implementing AI into their workflows. If you want to stay in the know, consider subscribing to this newsletter:

This 6 minute NotebookLM video is a visual summary of the post, check it out.

MIT just dropped a report showing that 95% of companies are getting absolutely nothing from their AI investments. Billions down the drain. But before you write off AI completely, let me tell you about the 5% that's actually winning.

Success story: There's this coffee shop in Canada called Brewed Awakening. They were constantly guessing what to order—either running out of stuff people wanted or tossing expired products. They started using TradeGecko's AI to predict demand based on past sales and local events. Simple as that. Now they waste 15% less food and run out of things 20% less often.

Epic fail: McDonald's spent three years trying to get AI to take drive-through orders. Turns out "I'll have a Big Mac with no pickles but extra sauce" is rocket science for robots. They gave up. Even McDonald's couldn't make this work.

Here's the kicker: only 40% of companies have official AI tools, but 90% of employees are already using ChatGPT at work. Your team's probably already ahead of your company's AI strategy.

PetPals is a small pet shop in the UK. Their problem? Generic marketing emails that nobody opened. Solution? Mailchimp's AI that sends personalized emails based on what people actually buy.

Buy a dog blanket? Get suggestions for matching beds and treats. Makes sense, right?

Results: 35% more people opened emails, 25% more sales. People started coming back more often because they were getting useful suggestions instead of random spam.

The lesson: Pick the biggest pain at the moment and fix it with an AI feature of the tool you already use (most have them now). No need to revolutionize, just start using the tools you already have smartly.

What it costs: Free to start, then around €26/month for most small businesses

Setup time: Literally one minute

The real deal: The basic AI features work great for simple stuff. Want to summarize emails or categorize customer requests? Perfect. But when I tried getting fancy with chained prompts and knowledge bases, it got messy fast.

I also tested their AI agents thing with Pipedrive integration. Total disaster—kept giving me fake data. But I’m pretty sure it’s due to the MCP Server itself because I ran into the same issue when I tried to implement a similar thing in Cursor.

Verdict: Worth it for most small businesses. Huge integration library, works in a day, and solves real problems. If you need something more complex, check out n8n, Make, or Gumloop.

Take all those survey responses, customer interviews, and feature requests gathering dust. Feed them to ChatGPT (or Claude, Gemini, whatever you use) and ask it to group them by theme and sentiment.

One grocery store found out customers really wanted more dairy-free options—something buried in hundreds of responses. They added more dairy-free products and customer satisfaction jumped 20%!

Time needed: 30 minutes to set up and get insights.