TL;DR
In this episode of The Good Thing, Stefan and Jens sort ten AI startups into a tier list ranging from unicorn potential to dead on arrival. They debate landing pages, flimsy value props, and whether OpenAI will make whole categories irrelevant. The result? A skeptical but revealing look at how much of today’s AI hype feels like a bubble.
A game of Startup Survival
The episode opens with a twist: instead of interviews or industry news, Stefan and Jens play a game. Their content producer compiled ten AI startups, and the hosts graded each one on a tier list.
The tiers
- S Tier – Unicorn potential
- A Tier – Will probably survive and find a market
- B Tier – Decent shot, some good signs
- C Tier – 50/50 territory, hard to call
- D Tier – We’re worried, but not fully out
- F Tier – RIP. Defunct imminent
Neither host had heard of most companies on the list, which made the exercise more about gut checks and product sense than prior research.
The Missing Problem Statements
One startup branded itself as “the AI testing agent for your software team.” But neither host could figure out what it actually solved.
Jens pointed out that many pitches relied on vague slogans like “with the power of AI,” which meant very little:
If your pitch is just ‘with the power of AI,’ what you’re telling me is you do something with computers. That’s the most boring story ever.
Without a clear articulation of a problem, the company landed in the RIP category. Several other startups faced the same critique, with the hosts returning often to the lack of strong problem statements.
Wrappers, Clones, and Déjà Vu
As they moved through the list, patterns emerged. Several startups looked like thin wrappers around existing tools or clones of better-known products.
Stefan compared the trend to the dot-com era:
This feels like the .com boom, when everyone just added .com to their name.
Both hosts agreed that many of these businesses would be easy prey for OpenAI, JetBrains, or VS Code once those platforms decided to expand into the same features.
When a Pitch Does Land
Not every company fell flat. The hosts gave credit to an AI sales role-play platform whose social proof, testimonials, and enterprise integrations looked more credible than most.
Jens noted the product at least addressed a real problem:
From my experience, you absolutely want to figure out your top reps, the behavior, and replicate this across the team. That’s great.
Even so, they stopped short of calling it a unicorn. At best, it landed in the “good chance to survive” tier.
Signs of a Bubble
By the end, none of the ten startups earned the unicorn label. Several were written off as RIP, while a few lingered in coin flip territory.
Stefan summed up the mood:
Yep, it’s a bubble. We are 100% in a bubble right now.
For both hosts, the exercise reinforced how much of the AI market is overcrowded with shallow ideas and me-too products.
Closing thoughts
The episode ended with a reminder that building startups is always hard. Even companies they dismissed might pivot and succeed. But the exercise highlighted the need for clarity, differentiation, and realism in AI pitches.
Feedback is a gift. Take it or leave it, but start with the actual problem you want to solve.
This episode was directed by Jacob Javor. Transcript lightly edited for clarity and flow.
