Founding and Scaling After ChatGPT: Building AI-First, Defending SaaS, and Getting Found in LLMs – with Henrik Roth (Serial Entrepreneur, ex HappySupport, ex neuroflash, ex BeWooden)

What Has Changed in Building Startups Since ChatGPT, Whether SaaS Still Holds Up, and How Visibility in LLMs with GEO and AEO Actually Works

In the twenty-first episode of the PRODUKTKRAFT Podcast, I talk to Henrik Roth. Henrik is a serial entrepreneur: he built BeWooden from zero into a seven-figure e-commerce business and sold it, then was co-founder and CMO of neuroflash, one of the best-known AI content platforms in the DACH region (Germany, Austria, Switzerland), and most recently founded his third company, HappySupport. A personal note on that: between our recording and the release of this episode, Henrik and HappySupport parted ways, and what comes next is still open. Our conversation is therefore also a snapshot from a phase in which the story hasn’t fully played out yet. To me, that makes it more valuable rather than less, because it’s precisely from this perspective that Henrik brings something rare in our scene: he looks at digital products as an entrepreneur, a marketer, and a product person at the same time. That’s why we start by looking at what this generalist role entails, where it helps enormously, and where it gets in your way.

Then we get into the topic that runs through the whole episode: how did founding a company work before AI, and how does it work today? What distinguishes an AI-first company from one that isn’t, and how can the transformation succeed? We get concrete and ask what can really be automated today. On the one hand in building, along the chain from interviews to opportunity solution trees to tickets and pull requests. On the other hand in the rest of the organization. Because the experience many people are having right now is this: the individual steps work surprisingly well, but every handoff falls short.

Then it gets uncomfortable. Is SaaS still worth it when customers can vibe code features themselves and rebuild everything you offer? Is the much-invoked SaaSpocalypse real, and how do you respond to it wisely? These questions lead to the next one: what’s new in product and business design? We talk about the feature factory and outcome thinking, and about which pricing models still hold up in this world, whether free, per seat, or outcome-based.

The most space goes to a topic where Henrik draws on his full experience as a practitioner: distribution after ChatGPT. I’ve already talked about distribution in general on this podcast with Carsten Bohling; today we take a deep dive into GEO and AEO. How do you approach it, and why might there be more potential in it against the big players than in classic SEO? We go through the building blocks one by one: prompt monitoring and the question of which prompts you need to measure in the first place and what volume lies behind them, fan-out, third-party validation via Reddit and co., which reminds me suspiciously of the link farms of the past, and E-E-A-T. And of course we discuss whether classic SEO still has a future.

Finally, it gets personal and tangible for your own work. Does growth bring the disciplines of founder, marketing, and product together? What do product people get wrong about entrepreneurs and marketers, what can they do better, and where would they need to work much more closely together? And the question that probably affects most of us: what can product managers who currently work in non-AI-native organizations do to keep up with the times?

Have a listen to the twenty-first episode of the PRODUKTKRAFT Podcast to find out how AI-first founding, modern product work, and visibility in LLMs really fit together today!