Stay organized with collections Save and categorize content based on your preferences.

Deepomatic: More Efficient and Sustainable with Google Cloud

About Deepomatic

Deepomatic is a French scale-up—part of 'FrenchTech' ecosystem—which develops an AI-based visual automation platform. It equips field workers, mostly in Telco and Utilities industries, to help technicians provide qualitative service and empower back-offices with monitoring capabilities.

Industries: Technology
Location: France

Tell us your challenge. We're here to help.

Contact us

A leader in AI image recognition algorithms, the scale-up Deepomatic chose Google Cloud in order to simplify its operations, accelerate its machine learning, and streamline its international expansion, all while favoring a sustainable approach.

Google Cloud results

  • Uses managed services offered by Google to focus on business and efficiency rather than infrastructure maintenance
  • Manages approximately 110 production servers, as well as more than 3,500 containers with only 2 DevOps
  • Cut data processing time with BigQuery: from 32 minutes to 7 seconds on a complex request

Benefits from significant flexibility to optimize costs and carbon impact

This FrenchTech gem created in 2014 has a host of accolades. In 2019, deepomatic was ranked among the world's Top 10 "Computer Vision" (image recognition) platforms by Forrester New Wave™. One year later, it was also labeled the "most promising European scaleup of the year" by the European network EIT Digital, supported by the European Commission. With an annual growth in turnover of 200%, the company is not only present in Europe and South America, but also in the United States. In addition, it counts some key market players among its client base, including Unit-T, Bouygues Telecom, Stellantis, Sanofi, or Movistar Colombia.

Focusing on added business value

For the last six years, this French company has been developing an artificial intelligence software platform that helps companies create and configure tailor-made image recognition solutions, without the need for programming or data science skills. Intended for quality assurance, its technology is based on a mobile app that allows field technicians to take photos at key moments during an intervention. These photos are then sent to the platform for analysis and, if the results are not compliant, new instructions are sent to the operator. "Performed in real time, this quality assurance increases the success rate of interventions and thus saves our customers considerable time by removing the need for them to go back and correct problems," explains Vincent Delaitre, Deepomatic's CTO. "It also makes it possible to gradually collect reliable data on the status of equipment in the field."

An employee doing field picture analysis, live feedback to the agent on the app, and an employee doing operational and quality analysis
Field agents get live feedback thanks to the ML-based analysis of pictures, which provides technical guidance and quality insights. Here is an example of cable box maintenance for the telecom industry. The pictures are analyzed in real time by Deepomatic to provide technicians with recommendations and perform quality assurance.

From the beginning, Deepomatic has opted for the cloud, relying on an architecture based on microservices and containers deployed on Docker Compose. The aim was to benefit from significant flexibility in terms of evolution. Kubernetes changed everything. "Our strength lies in developing innovative, added value features for our customers, not in maintaining the infrastructure to host our platform," Vincent Delaitre emphasizes. "From this perspective, Kubernetes was a true revolution. Where we once had to deploy and manage a virtual machine to host and maintain Docker Compose, Google Cloud offered us a fully managed solution with many more features. So, back in 2017, we switched to Google Kubernetes Engine in order to benefit from the best implementation offered by Kubernetes."

"From this perspective, Kubernetes was a true revolution. Where we once had to deploy and manage a virtual machine to host and maintain Docker Compose, Google Cloud offered us a fully managed solution with many more features. So, in 2017, we switched to GKE in order to benefit from the best implementation offered by Kubernetes."

Vincent Delaitre, CTO, Deepomatic

Capitalizing on managed services

Deepomatic doesn't really manage large load spikes, since its traffic is rather linear. However, the company doesn't want to waste time maintaining servers or configuring load balancing. "To support the diversification of our activities into new markets, as well as our international development, we also use autoscaling," explains Vincent Delaitre. "Cloud elasticity is clearly a huge asset for our strategy, but the range of managed services offered by Google Cloud is just as important. It allows us to not only focus on our business, but also be much more efficient."

Thanks to the automatic provisioning of resources managed by GKE, Deepomatic only needs two DevOps resources to manage approximately 110 production servers, as well as more than 3,500 containers. The company also plans to switch a large part of the processing performed by PostgreSQL to BigQuery. "Our first tests were very conclusive. Where it once took us 32 minutes with PostgreSQL to perform a complex request, it now only takes 7 seconds with BigQuery," continues Vincent Delaitre. "In general, we favor the services managed by Google Cloud whenever possible. Thanks to this approach, we now have a surprisingly substantial platform compared to the number of IT specialists we have, according to an investor who audited our system during our last fundraising campaign. This proves that Google Cloud significantly increases our efficiency."

Optimized resource management

Efficiency is not the only benefit provided by Google Cloud. Deepomatic is conscious of environmental issues, so it has developed a sustainable culture since its launch. By opting for Google Cloud, the company has also chosen a cloud partner that has been operating on 100% renewable energy since 2017, and one that is on a trajectory for net-zero emissions by 2030. The transparency and many tools provided by Google Cloud to help its customers in their carbon neutral approach are key for Deepomatic, whose cloud usage is the main source of emissions (around 80tCO2e/year in 2021). The flexibility provided by Google Cloud is also very important for Deepomatic's CTO. "While other providers impose servers on you without offering any wiggle room regarding the structure, Google Cloud allows you to allocate CPU or RAM according to your requirements. Both from an economic and ecological perspective, it's wasteful to provision resources that we don't need."

"While other providers impose machine sizing on you without offering any wiggle room regarding the structure, Google Cloud allows you to allocate CPU or RAM according to your requirements. Both from an economic and ecological perspective, it's wasteful to provision resources that we don't need."

Vincent Delaitre, CTO, Deepomatic

Another important point is that the computing loads for the algorithms developed by Deepomatic mainly use GPU resources, and Google Cloud offered the highest diversity in terms of graphics cards in 2017, when the company decided to migrate. "This is essential for us, because we need powerful graphics cards to train our models, but not to make predictions. In other words, this diversity allowed us to optimize our costs and our environmental footprint by allocating computing loads to different cards according to our actual needs," states Vincent Delaitre.

Aligned with the company's values, this efficient sustainable development policy provides a competitive advantage: faced with pressure from customers who care about their suppliers' environmental footprint, Deepomatic is already one step ahead.

Tell us your challenge. We're here to help.

Contact us

About Deepomatic

Deepomatic is a French scale-up—part of 'FrenchTech' ecosystem—which develops an AI-based visual automation platform. It equips field workers, mostly in Telco and Utilities industries, to help technicians provide qualitative service and empower back-offices with monitoring capabilities.

Industries: Technology
Location: France