Dear hustlers, founders, operators and visionaries,
Today’s guest is Andreas Goeldi, Partner at b2venture, who previously co-founded Pixability, a video advertising platform used by major global brands. He has worked in the internet industry since the early web era and now invests in and advises European startups while writing extensively about AI and software markets.
🎧 Tune in now on Spotify, Apple, YouTube and share your thoughts! In the meantime: Follow the Gradient and stay tuned!
PS: Has this e-mail been forwarded to you? Sign up here.
Why you should listen
You should listen to this if you want to understand why the traditional SaaS playbook is breaking in the age of AI and what replaces it.
As we kept diving deeper into the conversation, a clear tension emerged between fast AI demos that look magical and the deeper product and domain expertise required to build durable companies.
What we talk about
00:00 - Introduction
02:22 - The biggest delusion in AI right now
04:55 - Four buckets of AI business models
07:58 - AI pricing in times of vibe coding
13:55 - The real moats left in AI: data, regulation, and user experience
17:48 - Two startup ideas you should never build right now
22:17 - What existing SaaS founders must do to survive
25:49 - Why middle management is about to disappear
33:01 - How AI is transforming venture capital from the inside
37:11 - Do startups still need VCs when four people can hit millions?
43:03 - Rapid fire: thin wrappers, CTO hiring, and European quality
Our main take away’s
The SaaS playbook does not map cleanly onto AI businesses. Traditional SaaS assumed high gross margins and scalable software distribution, but AI introduces variable compute costs and pushes companies toward delivering outcomes rather than tools. This changes how products are packaged, sold, and priced.
Customers increasingly buy results instead of software. AI makes it possible to automate complex services while keeping the efficiency profile of software, which shifts products toward “service as software.” Goeldi saw this firsthand at Pixability, where AI automation allowed the company to more than double revenue in a year without hiring additional staff.
General AI agents will eliminate many shallow AI startups. If a workflow can be executed by a general agent integrated into tools like Excel or coding environments, a standalone product will struggle to survive. Startups need deep domain knowledge or highly specialized workflows that large model providers will not prioritize.
Defensibility shifts from features to context. AI makes it easier for competitors to replicate product features quickly, which weakens traditional software moats. Durable advantages increasingly come from proprietary data, regulated environments, and deeply integrated workflows that require real operational expertise.
AI adoption will reshape how companies are structured. Agents increase the productivity of individual contributors while automating coordination work that often sits in middle management layers. The likely result is flatter organizations with fewer coordination roles and individuals who manage AI systems rather than performing the tasks themselves.
How to reach out to Andreas
Exclusive from Andreas
If you were running a 100–200 person SaaS company today, what are the first three internal workflows you would aggressively automate with AI in the next 12 months?
Automating customer service without losing the personal touch for customers
Completely revisiting the software development process
Streamlining internal reporting and data access. There are a lot of hidden efficiency potentials buried in your data, and AI can help you find them.
In a world where AI makes building features much faster, how should founders think about product strategy so they don’t get trapped in endless feature competition?
Deeply understanding your customers is more important than ever. People don’t want to buy a convoluted collection of features, but a product that solves their problem elegantly and efficiently. And think ahead to what AI means for your customers and how you can best integrate with their own agent landscape that they’re probably already building.
What is one signal that tells you an AI startup has a real moat and one signal that tells you it’s just a temporary “thin wrapper”?
The type of customer references looks very different, and so does net revenue extension. Wrappers are often bought by customers who just need “something with AI” to check the box on this trend. This can lead to rapid revenue growth, but often churn rates and margins are terrible. Deep engagement that signals real differentiation shows up in numbers and customer statements.
How should founders rethink hiring in the AI era: what roles become more important and which ones will likely disappear first?
Any job that is just routine knowledge work, such as the application of standard templates to repeating problems, will disappear first. Look for people who are eager to use these new AI tools and are driven enough to reinvent themselves and their work.
If AI drastically lowers the cost of building products, what will actually become the hardest part of building a company in the next five years?
There are many theories about this currently, and I don’t think we know the answer. I suspect that reaching, converting and retaining customers will get much harder. But the new playbook is just being written right now. To me it feels like the early days of SaaS when nobody knew the answers for this new category.
What is success for you?
When I was a founder: Making customers happy and inspiring employees. Now as an investor: I enjoy seeing the teams we invest in grow and build amazing companies.
What books, podcasts, articles inspired you?
I actually have a whole website with tips (many true classics, although I should update it again): notion.so/agoeldi-s-Startup-Reading-List-73a477bd8de04929a5ebbd073fc78493 I find business books to be increasingly less relevant, but the current content you can find on Twitter/X and Substack is much more relevant in these turbulent times. My favorite podcast right now is probably Moonshots by Peter Diamandis.
What’s one advice, founders should actually ignore?
The most ineffective advice is when advisors or investors use analogies and recipes from their past experience without reflecting if the situation really applies to you in the same way. Always ask yourself if the person giving advice has the right context and knowledge about your situation. The very same piece of advice can either be brilliant or destructive depending on the context.
What are habits, activities or rituals that keep you sane?
At this point I have given up on sanity…
No, to be serious: Journaling is very important for me to reflect on current events. And it’s a useful habit to make predictions in your journal and then check back after a few months how things actually turned out. This gives you a strong feedback loop about your biases. It’s a technique described by the legendary management scholar Peter Drucker in his classic article “Managing Oneself”, which I highly recommend.
What is one “growth hack” that has a positive impact on you or the company?
Don’t make decisions without getting a deep enough concrete understanding of the relevant situation or object. For example, many executives make decisions about investing in some new technology without ever trying it themselves. This can lead to overestimating the relevance of a technology (e.g. the Metaverse) or underestimating it (e.g. AI agents for non-coding tasks). Being hands-on is the most important thing for founders and feeds your intuition. Demand the same from your employees.
Follow the Gradient is a weekly newsletter and podcast by the serial founders Melanie Gabriel & Christian Woese about how to scale a business from Europe while staying sane.
You have full control over what we send to you. We sent a weekly rundown on Tuesdays, and our podcast brief on Thursdays. If you do not want to receive this Thursday one, you can change this in the email preferences below.
