- Exploding Ideas
- Posts
- Reflection 1 - Exploding Idea
Happy Sunday,
This week we’re going to do something a bit different. Over the past few years we’ve been sending out this newsletter with new ideas, but hadn’t yet reflected on old ideas.
So in this Exploding Ideas we’re going to do a reflection. We’re going to look back at some ideas we previously covered and reflect on how they’re currently doing. We’ll look at 2 ideas that are continuing to grow, and 2 that bombed.
On Tuesday all “pro” subscribers will get the deep dive for the top growth idea.
👉Want access to the Ideas Database? Click here (link).
Read time: 1 min 24 sec
Exploding Idea #1: GLP-1
Why we chose it:
When we first covered GLP-1 drugs (Ozempic, Wegovy, Mounjaro), it was clear this wasn’t just another pharma story — it was the first real biological breakthrough in behavior modification. Weight loss drugs had always been scammy or short-term; GLP-1s finally produced sustained results with clinical backing and FDA approval.
Why it’s still booming:
This one earned its hype the hard way: it actually works. People are losing 15–20 % of body weight and keeping it off. That’s real clinical impact — not placebo, not influencer spin. The broader ripple effects have been wild: grocery chains are reporting drops in snack and soda sales, airlines are literally revising weight averages, and Big Food is quietly reformulating for “GLP-1-friendly” products.
Add in new studies linking it to heart-health and addiction recovery, and you’ve got a genuine multi-industry disruptor. It’s one of the rare ideas where the hype undershot the outcome.
Exploding Idea #2: RAG (Retrieval-Augmented Generation)
Why we chose it:
We saw RAG as a backbone of the AI infrastructure wave — the logic was: “LLMs hallucinate, so you plug them into your own data.” For a moment, it looked like every enterprise would rebuild their workflows around this pattern.
Why it’s still booming:
The chart shows slow, durable growth — no huge spike, but steady climb. That’s because it’s useful but invisible. RAG didn’t become the hot trend; it became the plumbing. Still, some of the hype cycles were fueled by LangChain and Claude ecosystem marketing — blog after blog calling it “the future of enterprise AI.” The reality: it’s expensive and hard to do well. Most “RAG products” devolved into glorified document search. The ones that stuck (Perplexity, ChatGPT Enterprise) just embedded RAG deeper into their stack. So while it’s not a pump-and-dump, it’s more of a solid infrastructure layer that never became the headline story.
Exploding Idea #3: Workflow Automation
Why we chose it:
We bet on “workflow automation” because it always surges when the economy tightens. AI or not, companies want to replace humans with scripts.
Why it’s bombing:
The trend exploded earlier this year — mostly because of viral posts about automating your job with ChatGPT and Zapier — and then crashed as reality set in. Most of those setups didn’t scale, broke constantly, or required technical upkeep most people didn’t have. There was also an AI fatigue backlash: too many “make money while you sleep” automation courses flooded X and TikTok. That spike you see? That’s the hustle-wave top. Now, it’s normalized back to baseline — the domain’s not dead, it just reverted to its real audience: ops managers, not creators chasing passive income.
Exploding Idea #4: Model Context Protocol (MCP)
Why we chose it:
When Anthropic announced MCP in late 2024, it looked like a big deal — an open “USB-C for AI” standard that would let apps and models talk to each other easily.
Why it’s bombing:
This chart is the definition of a dev-Twitter pump-and-dump. It spiked when the Claude team and early open-source devs amplified it as the next big thing. Everyone wanted to claim they were “building on MCP.” Then people tried it. It was overly technical, poorly documented, and lacked real adoption outside Anthropic’s ecosystem. No killer use case, no ecosystem support from OpenAI or Google, and the hype cratered within months. Right now, it’s mostly a ghost — mentioned occasionally in AI dev circles, but with no meaningful traction.
Learn how to monetize this trend in the Deep Dive (link here)