With 56% of CEOs reporting zero ROI from their AI investments, Matt Edmundson takes a refreshingly honest look at the four AI tools he actually uses across his ecommerce businesses right now. In this solo Slingshot episode of the eCommerce Podcast, Matt breaks down his monthly AI spend of roughly £350 and explains exactly how each tool fits into daily operations at Aurion, from deep research sessions to product photography and building what he describes as a digital second brain. Rather than chasing every shiny new tool, Matt shares how his team culled their AI subscriptions and settled on a focused toolkit that delivers real results. He also tackles the thorny issue of team adoption and offers a practical challenge for anyone still sitting on the AI fence.

Key Points:

1. Claude Code and Obsidian as a Second Brain [00:05:00]
2. Deep Research with Perplexity [00:13:00]
3. Learning Smarter with Google Notebook LM [00:16:00]
4. Making the Tools Work Together [00:23:00]

Claude Code and Obsidian as a Second Brain [00:05:00]

Matt’s primary AI tool is Claude on the Max plan at around £150 per month, and he pairs it with Obsidian, a note-taking app that stores everything as plain markdown text files on your computer rather than locking them away in someone else’s cloud. The real magic happens when Claude Code connects to this system.

“Think of the difference between texting a plumber for advice versus having the plumber in your house with their tools.”

That’s the difference between using a chatbot in a browser and running Claude Code in your computer’s terminal, where it can see your files, run commands, and make changes directly.

1. All company information, branding documents, playbooks, and scripts live inside one Obsidian vault
2. Claude reads thousands of notes and even learns and updates its own files over time
3. Matt describes the result as “more like having a team member who has spent six months reading every document you have ever written”
4. Everything stays local on your machine, which is a significant security advantage over cloud-only tools
5. The migration from his previous app (Craft) to Obsidian took about two days, and the system has been running for roughly three months

Deep Research with Perplexity [00:13:00]

For research tasks, Matt turns to Perplexity at around $20 per month. Unlike a standard chatbot, Perplexity provides sources with clickable links so you can verify everything it tells you.

1. The narrative binding episode (episode 274) came from a full-day Perplexity research session that produced a 30-page document
2. Matt uses the voice chat feature during his Wednesday morning walks, turning exercise time into research time
3. The sourced approach means you can trust and fact-check the output rather than blindly accepting AI-generated claims

Learning Smarter with Google Notebook LM [00:16:00]

Google Notebook LM, part of the Google Gemini suite at roughly $20 per month, takes a different approach to AI-assisted learning. Instead of drawing on the entire internet, it restricts its answers to the sources you upload, with a limit of up to 300.

1. Matt used it to study negotiation techniques, uploading both Getting to Yes and Never Split the Difference and then asking questions across both books
2. The audio generation feature creates 20-minute podcast-style conversations from your uploaded sources, making it easier to absorb material on the go
3. Because it only references what you give it, there’s far less risk of hallucinated information creeping in

Making the Tools Work Together [00:23:00]

The real value comes not from any single tool but from how they connect. Matt outlines a workflow where Perplexity handles the initial research, Claude Code turns that research into playbooks and frameworks, and those playbooks generate prompts for other tools like Nano Banana (Google Gemini’s image generation, used for product lifestyle shots).

1. Nano Banana has been used for product photography, including an Omega-3 bottle lifestyle shot with dolphins, though Matt still works with his photographer Lindy for key shoots
2. AI supplements the creative process rather than replacing it
3. The system stays current over time because Claude updates its own reference files as new information comes in
4. Team adoption has been gradual. Even the dev team were slow to pick it up. Not everyone needs the full setup, and the admin team use Claude with project files and specific prompts tailored to their roles
5. Matt’s challenge to listeners is simple but pointed. Pick one thing, give it a proper go for two weeks, and remember that AI is a co-pilot, not a replacement

Episode link: https://www.ecommerce-podcast.com/how-im-using-ai-in-my-ecommerce-businesses-right-now