In 2011, we began eFounders with Quentin Nickmans. The concept was to create several companies in parallel through a studio model. At that time, everything was to be invented; we had neither the means nor the team. We started with Mention, Mailjet, and Textmaster during the first two years. This allowed us to learn a great deal, such as our level of involvement, the type of entrepreneurs we wanted to partner with, and the method to be implemented. By 2015, we had launched a total of 5 companies, just over one per year on average.
It was only 5 years after our beginnings in 2016 that we began to expand our team (the Coreteam) with the goal of reaching 5 startups created per year. This new goal was combined with a 10 million euro funding round that allowed us to graduate to the next stage. Since then, we have maintained a pace of creating 4 to 5 startups per year. Twelve years later, over 40 startups have been created from scratch in our startup studio.
While part of the value created in recent years still remains on paper, I believe we can recognize the success of the model we created. We have created about forty companies; several have already been sold, most have raised funds, they generate together 450 million Euros in ARR, employ over 2500 people, and have created an estimated total value of 4.5 billion Euros.
We want to take this studio model we have imagined, this model we like to call team entrepreneurship, even further. What we did 7 years ago, accelerating from creating 1 to 5 companies per year, we want to replicate by going from 5 to 30 companies per year by 2030.
The challenge is huge, particularly because we have realized that our work is very artisanal by nature, creating companies requires a strong commitment, and we will not be able to create more companies per year even by expanding the size of the core team.
To grow and go further, we need to solve two problems - which is what led us to the creation of Hexa. First, we need to think about how to scale operations to increase the number of companies without losing the quality of operational support we provide. Second, we need to scale funding to continue to accelerate the creation of companies without being limited by our financial capabilities.
To solve this two-variable equation, we were inspired by a model we are often compared to yet are completely different from.
YCombinator - an inspiring scaling model
YC, created by Paul Graham in 2005, laid the foundations for the acceleration model by offering a 3-month support program and financing in exchange for 7% of the equity. They attract the best companies in the world and have established themselves as the leading accelerator. There's YC... and the others. Interestingly, our models are not incompatible since 7 of our companies have participated in Y Combinator.
We are often compared to Y Combinator because, like them, we offer "unfair advantages" to entrepreneurs to succeed in their businesses. Like them, we have democratized a model, them with the accelerator model, us with the studio model. However, one fundamental difference is that they look for projects, we look for entrepreneurs. It may seem like a detail, but it changes everything in terms of communications and value proposition.
We were inspired by YC’s 2014 phase: when they started scaling. YC went from a program of a dozen startups led by only one person, Paul Graham, to a much broader model that today welcomes almost 1000 startups per year. Despite criticisms, they have managed to grow while maintaining their values and level of demand. The different iterations of their scaling process is a huge source of inspiration for us but we sought to create something uniquely tailored to our studio's model.
Operationally, YC went from a single batch to a cohort system divided by theme, and each managed by one or more partners. Each batch is divided in several cohorts by theme and each cohort is lead by one or several partners.
At Hexa, we're broadening our scope from the specialized 'future of work' vertical to multiple verticals, each led by one or several Partners. These Partners are typically seasoned entrepreneurs or operators with extensive expertise. Opting for a collective approach over individual ventures, they will leverage a small team and Hexa's core resources to co-found 2 to 4 companies annually, earning direct equity compensation for each venture. For instance, Matthieu Vaxelaire now heads the 'future of work' vertical (the historic eFounders studio), while Florent Quinti is a Partner focusing on our web3 and AI verticals, while Camille Tyan concentrates on fintech. We're excited about introducing new verticals in health and education by the end of the year, and we’re looking for Partners to join us by either launching new verticals or contributing to our existing ones.
Additionally, we will soon be announcing a new initiative, which doesn’t fully fall within our startup studio model, but still focuses on lending our operational help to founders, in this case not to build with them but to grow with them, a strategy previously successful with Yousign.
To welcome more startups and all these new initiatives, we decided to move to larger, but more importantly, a better suited space for our way of working. Since the beginning of September, we have been located in La Cristallerie, a 800m2 space in the heart of Paris. We dream of making it a must-visit location for international visitors and one of the focal points of the European ecosystem.
Financially, YC relied on external investors to fund its startups from an early stage. Starting in 2015, Michael Bloomberg financed 100,000 EUR out of the 120,000 EUR invested in each startup, taking a 1% equity stake, while YC retained the remaining 6%. Currently, at Hexa, launching a startup costs us 800,000 EUR, accounting for both the direct expenses of the company and the studio's indirect costs. Although funding 4 to 5 startups per year was feasible, scaling up to dozens annually poses a significant financial challenge. To overcome this and not constrain our growth by our capacity for upfront investment, we've established a fund to finance our ventures. We’ve raised 20 million EUR with this inaugural fund. It will cover a substantial portion of the startups' initial costs, in exchange for consistently acquiring equity in all new ventures. This approach, which we conceived a decade ago, required us to first demonstrate our model's viability to secure investor confidence.
This is why we have just raised 20 million Euros, as part of a fund which we called the Hexa Build Fund. It will help us finance the whole building creation part of the process and allow them to emerge in the best conditions.
Thanks to our new organization with Partners and this new fund, we are now able to significantly scale up the number of companies we create each year, achieving our goal of increasing from 5 to 30 businesses. This approach costs us about half of the shares we used to hold in each created company, but it's by sharing that we ensure the sustainability and scalability of our model.
Nearly two years of crisis, perhaps the end?
I founded my first company, Fotolia, in 2004, coming in after the 2000 crash, and for 18 years I experienced the careless euphoria of a market with infinite growth. It seemed easy to me, just come up with a good idea, charm investors, and execute well to create exceptional companies. Everything that happened only confirmed this view, like Fotolia's sale for 800 million dollars, the success rate of our Hexa companies, and the three unicorns we created—Spendesk, Aircall, and Front—in just a few years.
Post-COVID, this growth accelerated to an unhealthy, almost unbelievable level. Then came 2022, with poor results from publicly traded tech companies dragging down the rest of the market. More than 18 months later, we're still there, with a sluggish ecosystem, consistently disappointing quarterly results, complicated to impossible funding rounds, and a very real pervasive gloom.
This shock proved beneficial as it reset the clock and forced everyone to face reality. We've shifted back to creating business plans, prioritizing profitability, and exercising caution before investing in any PowerPoint presentation.
At Hexa, we've been hit hard too. All the companies in our portfolio suffered, the biggest ones like Aircall, Front, and Spendesk, which had thrived in recent years by raising amounts that elevated them to unicorn status. For the past 18 months, they've had to lower their ambitions and adapt their growth plans by reducing their scope or goals. Timely raised funds have significantly limited the damage.
Regarding our studio and company creation, the crisis we're going through hasn't stopped the idea machine. We've never created as many companies in a year; we've launched six since the start of the year and plan to launch at least two more by the end of 2023. It would be an incredible feat.
However, it has never been so complicated for our companies to raise money during the last 12/18 months, while it's generally much simpler in a studio—this indicates even greater complexity for others... Our business model necessitates that our companies secure funding to achieve independence. Consequently, we are adapting our strategies to present VCs with companies that demonstrate real revenue instead of just traction, and have profitability objectives rather than unchecked growth ambitions. Adapting to this new paradigm is essential for the continued success of our model.
The concept of "Design Partners" has replaced that of "Pilots," the notion of "Sales Pipe" has replaced "Waiting List," and the concept of MRR (Monthly Recurring Revenue) now takes precedence over user numbers.
Some signs suggest that a semblance of trust is returning, that we may have hit the bottom, and that better days could be promised. As long as the light at the end of the tunnel isn't brighter, we must maintain a conservative posture, which has the advantage of pushing us towards greater pragmatism.
It's surprising, for someone who has only known rapid growth and who long believed tech was a world apart, to realize that we create companies (almost) like any other, with real economic constraints—and that's not necessarily a bad thing.
The AI shock, a new world?
As the tech market faces its darkest hours, a bombshell has dropped in our world. Something unexpected, the most impressive thing I've ever seen.
End of 2022, just over a year ago, the first version of ChatGPT was released. It took most of us just a few minutes to realize that it was a revolution. The quality of the responses was exceptional, a hundred times beyond what we had seen or imagined in our wildest dreams.
Even the most expert and optimistic had not predicted such rapid evolution of the models. Although, as usual, it's a combination of factors like the ability to manage larger volumes of data or increased computing power that have allowed the development of these technologies, it's the transformer architecture and attention mechanisms directly from research that have propelled artificial intelligences. The small lab teams working on these topics have won; they have come out of the shadows.
It's all the more striking because it's unexpected, and the progress since then has been extraordinary. Since then, a lot of grey matter, technology, human and financial resources have been invested in the accelerated development of these new technologies.
Here, obviously, in the world of software and SaaS, there's a bit of an upheaval. We can feel that AI is going to change everything, meaning we've grasped its impact, but we don't yet know in which direction or how to tackle this revolution. For instance, all the new ideas we had at the end of 2022 have all withered away. There is an overabundance of ideas, as evidenced by the batches of Y Combinator filled to more than half with companies centered around Generative AI. It is not easy, at the beginning of a new era, amidst tremendous noise, to find what to focus on and understand where to direct our attention. To think calmly, we need to sit on pillars, strong convictions that we have and that will allow us to separate the wheat from the chaff.
1 - Interfaces: Artificial intelligences need a framework to fully express themselves and to be used concretely by users. The "chat" format we know today is a primitive form of interface that has the merit of being easily accessible but does not allow for sophisticated tasks. Just as an AI for an autonomous car would make absolutely no sense without the car attached to it, each intelligence will need a dedicated interface to be fully consumed. AI will transform software, not replace it. We will always need specific interfaces for note-taking, writing, designing, drawing, organizing, selling, calculating, etc.
2 - Specialization: Today's LLMs like ChatGPT, Bert, and Co. are large generalist models. They are impressive at first glance (a bit like when you meet someone very cultured on all topics) but they are extremely demanding in terms of space and performance. We can imagine that the future will be made up of much lighter and more specialized models, even integrating dedicated interfaces as mentioned. For example, we can envision a model to help you create your low-fidelity wireframes on Figma or a model to translate medieval Germanic texts.
3 - Distribution: Transformer technologies and attention mechanisms are now known to everyone. If ChatGPT surprised everyone, it is today the state of the art in technological terms, and most other models, including open-source ones, will soon have the same performance. There will thus be a commoditization of generic models. Value creation will certainly come from model specialization through a set of data and a proprietary interface. For generic models, it will be distribution that makes the difference, and in this game, Google, Meta, Microsoft, and the new entrant OpenAI will be the ones to stand out. An opportunity may lie in the ability to find alternative distribution models such as OEM or On-premise, for example.
4 - Digital Twin: I remember the beginnings of Web 2.0. Everyone was trying out collaborative models, taking the old models and adding a contributory system (me first with Fotolia, which was just the Web 2.0 version of Getty Images). Then the social networking model arrived and swept everything away with it. As social networking was the culmination of Web 2.0, I have the intuition that the culmination of this new era, at least at the consumer level, will be that of the digital twin. That is, the digital self, a digital version of ourselves in a world where our presence is required in two places and where we still do not have the gift of ubiquity. I'm not quite sure what to do with this pillar, but it could clearly have repercussions in the B2B world.
There was the world of software before, there will be the world of software after. Firmly grounded on these four pillars, we will begin to imagine what the future of software might look like in this new paradigm. And we already have a vertical dedicated to AI with two startup ideas.
Our transition from eFounders to Hexa, our ambition to go from creating 5 companies per year to 30 by 2030, the move to a multi-studio model is taking place in a new world, weighed down by a crisis that could last and boosted by new technologies that could upset everything.
We remain very optimistic. Our studio model and even more our new multi-vertical structure offer us incredible agility and modality.
We could not dream of a better context to end 2023 on a high note and embrace 2024, a year that feels as exciting as it is challenging!
Read the past letters below:
2022 - Hexa Letter #1: A new way for entrepreneurship
2021 - eFounders Letter #9: 10 years of Team Entrepreneurship
2020 - eFounders Letter #8: Scaling the Startup Studio Model
2019 - eFounders Letter #7 — Cracking the Startup Studio model
2018 - eFounders Letter #6 — A Software Galaxy
2017 - eFounders Letter #5 — a Startup Studio on a Mission
2016 - eFounders Letter #4 — Time to touch base
2015 - eFounders Letter #3 — Rise of a startup studio
2014 - eFounders Letter #2 - Birth of a startup studio
2013 - eFounders Letter #1: DNA of a Startup Studio