My '10x Founder' Experiment: Building a Product in 9 days with Gemini
July 2025
by Miguel Santos [Mike] LUPARELLI MATHIEU
1225 is the number of papers, white papers, reports, insights, or any other kind of document I store in my “knowledge” folder. More than 500 tweets posted during the last two years. Not because I’m a big fan, but because I’m using it like my knowledge base of news and tech reviews.
The PROBLEM
Every time I have to prepare for a speech, a presentation, or writing an article I kind of struggle to find among many subjects like Economics, Geopolitics, Strategy, Data Science, Machine Learning, Artificial Intelligence, Quantum Computing, Blockchain, Cryptography, Biometrics, Digital Identity, etc., the right intel, and though my knowledge base is well organized, there is always a valuable amount of time dedicated to the curation of that knowledge. Picking and processing that information, and extracting the most relevant and insightful information takes time. That’s the problem to solve.
The IDEATION
Inspired by the 10x Founder concept introduced by Jeffrey Bussgang in his book The Experimentation Machine, I decided to try it by myself, i.e. leverage on Generative AI to build something quick. Because this project is part of my own experimentation machine (i.e. kind of a hobby) and out of the scope of my daily duties, I had a constraint to follow: 9 vacation days. That was a perfect constraint to try for real the potential of the 10x Founder thesis.
Note: Though the “ideation” process was a long journey (i.e. something that was around my thoughts for many months before), the whole process of building a product (Android app + product definition + website + go to market strategy) couldn’t be longer than those 9 days. Spoiler: I did it! Here is my story.
The EXPERIMENTATION (part I)
First, decide which Gen AI to use. That part was very easy. According to my previous experience and a few tries and errors, I chose Gemini to join me on that journey. Second, which environment to use, Apple or Android? I’m a big fan of Apple, nevertheless Android resulted to be more suitable for an on-device AI implementation. Though it has been more than 8 years since the last time I coded any line in Android, I manage to catch up quickly thanks to Gemini. It took me 7 days to complete the app. The remaining days were used to build the product definition, use cases, the website, and polishing some concepts and app UX/UI.
The PRODUCT (tech)
Augmented Digital Twin is an on-device (Android) knowledge base, a RAG (Retrieval-Augmented Generation), and Generative AI (Gemma). The user feeds the app with the knowledge they want to use for augmenting their insights capabilities. The on-device Gen AI (Gemma) guides the user through that process. Though it still requires some fine tuning (i.e. some prompt and chunk engineering to deal with on-device Gen AI limitations), it does the job of talking to the user bringing insights, key takeaways, and description taken from the knowledge base built by the user.
The EXPERIMENTATION (part II)
Differential thinking, pitch building, source aggregator, publication dispatcher, and team collaboration are some of its use cases and features that the Augmented Digital Twin can provide. Gemini (2.5 pro) was my partner along the way. First, acting as my co-DEV coding and bringing an explanation of each bug we found and solved. I mean, Gemini not only gives you a solution, but also an explanation. I learned a lot about the underlying architecture of Android during this process. Second, Gemini acted as my co-CEO defining the product, the use cases, and showcasing the differential features. Finally, Gemini adopted the co-MKT role to suggest a go to market strategy.
The LAUNCH
Gemini provided planning for recruiting testers, and picking a wedge use case to focus all efforts for the go-to-market strategy, i.e. "Go-to-Market Wedge" Strategy: "Effortless speech preparation for speakers." and a vision: "A fully integrated augmented knowledge base for professionals." Other dimensions of the business model were discussed, including price strategy, other use cases, but the priority is to test and collect feedback from this first wedge use case.
The LEARNING
I started this project with confidence and with uncertainties at the same time. Though I have theoretically acknowledged many of the benefits of working with AI, I wanted to test Jeff Bussgang thesis, and try by myself how powerful and transformative can AI be for Product Innovation. It was, somehow, a reskilling project. The result is outstanding! Mindblowing for someone who had struggled countless days and months solving bugs and building tech products in the previous era where Gen AI was marginal (i.e. GANs). Just thinking how easy it could have been to build those products…
Luckily for those who are product innovators, we are definitively in the era of augmented capabilities with AI.
IMPORTANT
I’m not saying that AI will replace all of us, because we wouldn’t be able to talk to Gemini as a co-CEO without having experience and a strategic mindset, nor co-coded with Gemini without knowing anything about coding and the underlying abstraction, nor talking about MKT without previous experience building go-to-market strategies. I’m saying that AI is augmenting our capabilities and accelerating the cycles of product innovation.