French startup Deepomatic has raised a $10.5 million (€10 million) Series B funding round. Although the seed round is relatively small, the startup has managed to convince some big customers to use its visual automation platform. For example, telecommunications companies use Deepomatic in the field to verify that tasks have been completed successfully.
EnBW New Ventures and Orbia Ventures are leading the newly announced funding round that Deepomatic closed in October. Existing investors Alven, Hi-Inov Dentressangl and Swisscom Ventures are again participating in the new round.
The startup has been around for a few years when I first covered Deepomatic in 2015. The company has always been focused on deep learning for computer vision applications. The main problem is that it was a long road to find the right customers for this technology.
With the telecom industry, Deepomatic seems to have finally unleashed its true potential. “We discovered an industry that really needed what we were working on—and that was telecommunications companies,” co-founder and CEO Augustin Marty told me.
When a field worker installs fiber optic cables or sets up a new 5G tower, they have to fill out complex forms to make sure they’ve followed some specific procedures. It can be quite boring as workers can work for contract companies. These companies may work with multiple telecom companies with different requirements.
It is also easy to make mistakes when filling out the form. Sometimes field workers can also tell that something is working well when it somehow works. It can cause some quality assurance issues, as we saw with fiber concentration points.
That’s why many field service companies also work with photos. After they finish the installation, they must take a photo of their installation and instruments to prove that some of the new equipment is up and running with the right parameters. It means more work.
At Deepomatic, field service companies mostly use photos as their yardstick. Photos are automatically analyzed to gain some knowledge. Deepomatic can then send some alerts if something is wrong and needs to be checked again.
“We started with the most difficult part, which is identifying the errors,” Marty said. In addition, Deepomatic now sells an end-to-end platform so that workers in the field only need to use Deepomatic to get something done. It also integrates with specific enterprise tools such as ERP.
When the startup is working with a new client, there is some integration work so that Deepomatic works exactly as expected. It includes adding control points, reusing some existing tasks in the computer vision library, or training the algorithm on a new set of photos. Deepomatic algorithms are trained on the startup’s own infrastructure. But its product can run on a customer’s own cloud infrastructure and, in some cases, on-premises.
The company currently has around 20 large accounts such as Bouygues Telecom, Swisscom and Movistar, as well as a bunch of smaller clients. Because it is business software, customers typically pay hundreds of thousands of euros per year to use Deepomatic.
Every month, Deepomatic monitors more than one million operations in the field. Every day, more than 20,000 field workers take photos with their phones and upload them to the Deepomatic backend.
Deepomatic and its team of 70 employees want to enter new markets and new industries such as renewable energy, electric mobility, construction, insurance, etc. Deepomatic wants to work with companies in Europe, USA and South America.
Many governments and large companies are currently investing heavily in renovating their infrastructure for the next few decades. At the same time, there is a shortage of talent for field workers. Deepomatic seems to be coming to market at the right time to become an essential tool for this infrastructure overhaul.