A true serial start-upper, Bertrand Guiheneuf has drifted from Inria to Fotopedia, crossing the Atlantic in order to breathe, in particular, the air of Cupertino.
PhD student in applied mathematics
First start-up, Henzaï, a Linux user interface for smartphones
Engineer project leader at Apple
Co-founder and technical director of Fotopedia
Technical director of the consultancy firm Hirondelle
From research to development
I was quite partial to technology, following the arrival of electronic toys during the 1980s, but I had never imagined you could make a career out of it! I did quite conventional studies in mathematics but, before I joined Inria, I had never done any computer science in a serious way. In 1995, wavelets - a mathematical object discovered, in part, in France - were having a major impact. Yves Meyer, one of the cutting-edge researchers at the time, was giving lessons at Dauphine university in Paris on this subject. At that time I was in Nantes, and I read his books, impressed by his teaching skills and his openness to applications. And so I found what I wanted to do: applied maths! I went up to Paris to do my postgraduate diploma (DEA) with him, and he suggested that I do a thesis as an extension of my studies. He was the one who, understanding my desire for applications, sent me towards Inria. The institute had a project, Fractales, led by Jacques Levy Véhel in Rocquencourt: I was a PhD student there on 2-microlocal analysis.
I knew a bit about the world of research through my father, who was a chemist: this world seemed a bit inflexible to me. At Inria, it was the opposite: there was a surprising freedom of expression and action! The types of 'mandarinate' I had heard about were totally non-existent at the institute. I did not experience the slightly hippy period at Bâtiment 8 that Jean-Marie Hullot can describe, but I did experience the boom in technology transfer. Our model truly was the MIT. Several Inria start-ups had been successful, others were just setting out. It was obvious that the main output choice should be industry. Moreover, I realised that the world of fundamental research, with its own legitimate metrics (references, publications, etc.) was not concrete enough for me.
Whilst technology transfer was in full swing at Inria, the world experienced a true revolution during the first part of the 1990s: open source. A new operating system, Linux, emerged, which was a very rare event. It was enormously successful and the system became predominant in the Unix world, in particular at Inria. At the institute, the idea of contributing to the development of free software became very important, in the form of software made by Inria or external collaborations. During my thesis, I familiarised myself with the open source contribution, and particularly on a software called Scilab, which has since led to the creation of a start-up. As I result, I moved from the research side to the development side.
Men and machines
In the open source 'evangelisation' plan, the obvious flaw was the human-computer interface. Everything worked really well when we were developing IT technologies we - those who were writing the code - were using, but when it was a matter of producing software aimed at users who were not creating them, everything was of quite poor quality. For example, accessing emails in Linux was relatively complicated: I tackled this problem and in 1999 was noticed by start-up creators in Boston, Ximian (since bought by Microsoft). I did some development there, which was interesting but slightly lacking in creativity! Georges Penalver, who was working on mobile phones at Sagem, was looking for people who were specialised in open source. He had a truly avant-garde vision: he thought that mobile phones would become computers and that they would require operating systems. Windows was not sufficiently high-performance and was very expensive, so another one had to be created. The resolutions weren't bad, but were unsuitable for a telephone model, so that also had to be improved. Georges Penalver proposed that I set up a start-up that would create this operating system using Linux: financed by Sagem prior to the launch, I got my team together...and, six months later, there was a restructuring at Sagem and our funding was cut!
We were in the process of building our prototype, we were sure that the project was a valid one and so we created our own start-up. Chahab Nastar, formerly Inria, introduced us to the inventor of the human-computer interface at NeXT, Jean-Marie Hullot, who was financing a few start-ups and working as CTO in an incubator. He agreed to back us, to introduce us to some business angels, etc. We hit it off really quickly, but then, once again, bang wallop! The Internet bubble exploded in 2000. All of the start-ups were beginning to close, we ourselves struggled to raise funding. So Jean-Marie considered joining Apple and proposed that I set up his team. I hesitated for five seconds.
A telephone at Apple?
In April 2001 we became consultants, then engineers, at Apple with the idea - which for some, at the time, was absurd - of making a telephone. It was a top secret project. The parent company struggled with this idea, while the United States were really not at the forefront on the subject: we needed to set the stage for it. For Jean-Marie Hullot, data was the key and, in particular, the synchronisation of these data and their availability on the different media. At that time, what data was there to synchronise on telephones? The address book and the calendar. On Macs, Apple had neither one nor the other. Everything had to be created. Initially, I mostly dealt with producing the calendar on Mac. However, from the very start, the idea was to use it on a telephone, so we needed to pre-empt the representation of a calendar on a smartphone.
Apple was quite a 'special' company: concentrated in the Cupertino district, closed, and with a cult of secrecy which, moreover, produced good results in terms of innovation. Apple has only minimal links with the rest of the industry and focuses on its consumers. The simple fact that an R&D centre created in France could produce software for Mac was far from straightforward. Fortunately, Jean-Marie Hullot had been CTO at NeXT and knew Steve Jobs very well. Despite everything, we had to prove our expertise. We worked like crazy and the alpha version was released in December 2000, after four relentless months. It is still in use today.
With human-computer interfaces, the work is excellent when nobody notices it! And, indeed, it was the first time that an event could be created so easily in a calendar. The keywords were rapidity and ease of access; they are still one of the guarantees of a successful application today. From 2003 to 2006, I was head of a team that was meant to lay the foundation stones of the future iPhone: the work focused strongly on content and the camera in particular. The iPhone development itself ultimately having to take place in the United States our team, which had bonded around Jean-Marie Hullot and preferred to stay in France, therefore set off for other adventures.
Documenting the world
We started Fotopedia in 2007, the project was financed in 2008 and developed until 2014. I am a technological person and I believe in the possibility of creating products with very complex engines and extremely simple interfaces. How could we turn cutting-edge technologies into products for everyday use? Jean-Marie Hullot was interested in photography, and we had the idea - just as Wikipedia and Google Maps were coming onto the scene - to document the world in the open source way thanks to photography.
So Fotopedia made it possible to put your photos online, to write captions offline, and to put them online when it was possible. The application led to collaborative heritage books - to show the world, explain the world, make it accessible and make it collaborative. Fotopedia created well-documented, structured data. Hidden behind this extreme simplicity at the service of the photographers was an unconventional collaboration technology - the result of the expertise acquired when we designed the iPhone synchronisation engines.
The life of start-ups is sinusoidal: it inevitably goes through doubts and questioning of the model (intellectual and economical). Our investors suggested that we open our product up to a wider public and we changed tack for a first time. We created a collaborative website with the aim of creating a collaborative, photographic encyclopaedia. The website was a success, and amassed a lot of quality content: for each Wikipedia article there was a corresponding Fotopedia document. The contributor community was very strong and extremely attentive to the quality of the photograph itself, however the daily traffic and the number of clicks were insufficient to guarantee financial income. Photography and challenging documentation are, in any case, difficult markets.
Apple brought out the iPad - which at the time had no application - and we changed direction a second time. We stepped into the breach! We had a perfect knowledge of the Apple environment, and we proposed a partnership to UNESCO in order to create an application on world heritage in photos, in the form of a very well-documented electronic book on each classified site. Surprise! The Apple App Store editorial team saw the application - without our having contacted them - and immediately showcased it. We went from 10,000 to 300,000 users, then to several million over a very short period of time. We reached financial stability thanks to advertising, in particular mobile application advertising which, for us, had a very good conversion rate into actual purchases. The advertising business model was then given a really rough time by Facebook, who undercharged for its inserts, and in 2014 we closed Fotopedia.
My first digital memory
For her 16th birthday, my sister was given a calculator on which it was possible to do programming in BASIC. My father thought that it was important that she learn programming: she has never done any but, for my part, I spent a year playing with it. It was so magical to be able to do things to a machine, ask it questions and get answers to them.
After eight years everybody wanted to do something new, so we each left in different directions. For my part, my passion is start-ups, but I wanted to tackle the unknown - and in particular the French ecosystem. I set up a consulting company, Hirondelle, which helps groups and start-ups to start off innovation projects at all stages (prototype, development, product, team and role management, business models, etc.). My goal, in fact, is to meet the team I will create my next start-up with!
In 50 years' time?
The impact of the voice in human-computer interfaces will be key. There has already been a spectacular increase in the quality of voice recognition, but the development of conversational chatbots remains basic. The introduction of real intelligent interactions will revolutionise the way in which we design human-computer 'interfaces' and, in collaboration with mobile tools, will erase the boundary between the material world and the digital world.