Synaplus : Democratize and accelerate AI adoption
Important KPIs ? On a day-to-day basis, Michel Bastide tracks revenue as a KPI to highlight because deeptech is growing on equity. Synaplus has been working in partnership with clients since the beginning to finance itself, and the second KPI to mention is the number of new clients per year; there are two important things when you create a software, it is both to keep your clients (factor of trust and recurrence) and to gain new ones (factor of attractiveness and development).
Any advice for Class #14 at the École Polytechnique incubator ? According to Michel Bastide, the most important thing as CEO of a deeptech is to listen to the market and adapt the strategy to the needs of this market (pivot, mutation). This listening to the market must be accompanied by a lot of lucidity; you have to constantly question yourself and your skills as well as the product or service you offer. In the end, the market will not be the one initially expected. Therefore, the complementary quality is tenacity; you must insist and not give up: "your product or service does not have the impact or the acceptance that you hoped for. You must constantly adjust, you must mutate, mutate, mutate...".
When this product or service fit with the market is proven (for a deeptech this can take 8 years as for Synaplus), it will be important to find the resources to finance the conquest.
In France, in the case of a deeptech that does not come from academic research, the difficulty will be financing the long period of R&D and market validation, and the risk of diluting the capital too early (because it is poorly valued at this stage).
DEEPTECH : SYNAPLUS
Synaplus was created in the very beginning of 2014 by Charles-Antoine Giuliani, Founding President & CTO, an alumnus of the Ecole Polytechnique (X94). Today the team size is 10 people and is looking to recruit :
- 1 Business Manager
- 2 Business Developers
- 2 to 4 Data Science Consultants
At the same time, they are preparing to raise funds after obtaining market validation through the acquisition of large customers, as the ambition is to conquer the global market. Synaplus is mainly competing with American companies, technologically less advanced but with a valuation between 4 and 40 billion euros.
"Our vision in 2 years is to structure Synaplus to bring it to a global stage".
The start-up's key objectives are :
- The conquest of the European market in the next two years
- In 2024, Synaplus plans to raise a second round of financing to conquer North America
Furthermore, Synaplus is part of the EIC Accelerator (step 1), an ambitious European funding program for innovative companies.
THE SYNAPLUS PRODUCT: COsMO Engine
AI can optimize all business processes; each company has a potential of several hundred AI projects to run. To do so, companies will have to hire hundreds of data scientists in the short or medium term; an unsustainable situation from a financial and operational point of view. This is one of the reasons why Synaplus has undertaken to develop COsMO Engine, the software that automates AI with AI.
"COsMO Engine is the AI engine supporting all operations of your data science project".
After 8 years of R&D, the team has created an AI operating system and COsMO Engine software that builds on it, to automate just about any business process from the data that customers have. This allows a data science project to be completed in days or weeks instead of months or years. And it can be used without being an expert; because COSMO Engine is a no-code solution.
"You don't have to clean the data anymore, the machine does that. You don't have to aggregate the data, you don't have to know the algorithms, the machine chooses them and trains them. The human skill is focused on understanding the business needs, on the data to be used, and above all on validating the results provided by COsMO Engine.
This validation is all the more valuable as Synaplus has succeeded in making it even more transparent than current practices: not only is COsMO much faster, but it traces and reports on all its calculations and operations, and it is able to present the results with a very high level of justification (feature importance by observation, variable simulation, model stability, etc.) in order to facilitate human validation.
"We have managed to develop the Excel of AI (COsMO Engine) and the Windows of AI (AioS) and this has a huge potential".
For the next few years, the team will continue to do R&D with the aim of developing the architecture to process all types of data in a company (text, images, videos...) in a correlated way and in real time.
It is the time between the moment when a customer tells them "I have a need" and the finalization of the datascience project ready to be exploited in production. The duration of this type of project is normally between 3 months and 2 years. Synaplus is therefore positioned as an important changemaker in AI. This does not exist on the market, it is a revolution!
Each week (for the first 6), the objective is to refine the client's business needs and, based on the data they have retrieved, they show the client if they can find an answer to their questions in the data. To do this, a workshop/interview is held one hour per week.
In the last week, they package the tool so that the client can use it in production; if the client wants an API integration, it takes another month to develop and format it on the client side (no development on the Synaplus side).
This 7-week project process is part of Synaplus' service offering and is called Proof of Value (PoV). The other services are :
- Ideation workshop. It is about contributing to the acculturation of the market, by helping the customer to identify the potential of AI and its applications in its business context.
- Production follow-up. This is the step that follows the realization of the datascience project, and that allows to follow the "AI product" in production (MCO, model drift management, re-training, change request management) in SaaS or On Premise.
- AI Program. Multi-year AI transformation program, allowing the customer to implement many AI projects in production per year and to become autonomous (fixed commitment of X projects in production per year for a fixed budget).
The first reason for joining X-TECH in 2021 is to have a scientific council with personalities from major laboratories such as INRIA or Télécom Paris. The presence on the Paris-Saclay Plateau brings Synaplus closer to institutional research, which is always willing to cooperate in order to bring new technologies.
Secondly, the X-TECH acceleration scheme allowed them to meet funds/investors during the Demo Day in December 2022.
Finally, Synaplus wants to be part of the Paris-Saclay ecosystem in terms of DNA, image and storytelling for the future.
Industry trends ? The hot topic is the standardization of AI, realizing that AI is going to be involved everywhere. At the same time, there is a big trend related to controlling the uses of AI, what we call ethical AI. In particular in health, fake news... The last and more important trend is about the trust we can put in AI.
"In fact, we don't realize it, but in terms of revolution, AI is more important than the appearance of fire or electricity - dixit the president of Google. Except that we don't see it yet and we don't understand it. We are in the prehistory of AI".
La France Digitale is closely following all the work and even advises on the laws that are in the works for the AI Act (in preparation). The goal is to make it evolve in the right direction for French start-ups & deeptech in the field.
"We are moving from AI, a luxury product, to AI, a commodity. Synaplus is helping to democratize AI".
The ecosystem ? The most important annual trade shows for Synaplus are WAICF, which takes place in Cannes in February, and Global Industrie, which takes place in Paris in May.