AI Experience Roadshow Europe 2019: Highlights from Stockholm, Paris, and Madrid

January 9th, 2020

This article is by James Lawson and originally appeared on the DataRobot Blog here:


Following the success of AI Experience London earlier in the year and in response to customer demand, we decided to extend the roadshow to other European cities. We held inaugural AI Experience events in Stockholm, Paris, and Madrid, attracting more than 300 attendees. Participants were highly engaged, evidently eager to start new AI projects or increase their capabilities.

Across the three cities, we followed a consistent format. Rather than doing a DataRobot sales pitch, we focused more on AI education. Our starting point was to consider the European Macroeconomics context - most counties have been struggling with productivity over the last decade. AI provides a special opportunity to boost productivity again, enabling people to have more fulfilling jobs and to deliver better products or services for customers.

Most organizations want to implement AI, but many find it challenging. We confronted this head on in the keynote presentations, discussing the opportunities created by AI whilst recognising that model building, deployment and monitoring can all be painful. To overcome these challenges, a wide range of AI capabilities are needed, not just Automated Machine Learning, but also MLOps and governance. Softer organization changes are also crucial, like pragmatic AI education (this doesn’t mean keeping people informed about the latest python frameworks, but equipping people across the organization to understand AI, identify opportunities, and frame new projects correctly).

Across Europe, attendees were also able to see demonstrations of DataRobot, with our team (Oskar Eriksson in Stockholm, André Balleyguier in France and Federico Castanedo in Spain) quickly building machine learning models live to show the basic capabilities of our platform. Enhanced functionalities from recent a release were also covered such as the AI Catalogue, the next generation of Automated Feature Engineering, and MLOps.

The biggest highlight of the roadshow were the sessions in each city with customers and partners sharing their insights. We’re particularly grateful for these contributions, as those executives are very busy and usually have to get special permission to speak so openly about their AI programmes. The result though is worth the effort, as participants gained a much deeper understanding of various use cases and the practicalities of successfully implementing AI - cutting through the marketing hype.

Stockholm Panel

In Stockholm, I hosted a panel with Vanessa Eriksson of Nets Group, Sudharshan Ravi of KPMG, Egil Martinsson of Schibsted, Tomas Beckeman of Deloitte, and our very own Director of Software Engineering, Brian Bell. In this panel, we heard about making better yoghurt, cutting fraud, recommending tags for news articles, email classification combining AI with RPA, and enhancing forecasting in retail. One customer shared that “it was a no-brainer to invest in DataRobot” given the cost and speed of traditional data science projects.

The topic of AI Ethics and building trust also prompted a lot of discussion. This prompted a panelist to suggest that, “every AI model should be explainable and show value ... managers needs to learn and not leave the ethical decisions to technologists alone”. This aligns closely to our perspective at DataRobot in our recent AI Ethics whitepaper.


Arnaud Bellétoile who leads the Data team at Cdiscount shared their experiences after a year of using the DataRobot AI platform. Cdiscount is the leading e-commerce site in France, with 40 million products and 9 million active customers. Their business was digital by default, with wide ranging AI use cases from marketing through to logistics. They naturally saw the value of AI, with a sizeable team of data scientists and supporting specialists. For them, DataRobot’s value was in accelerating the building and deployment of AI, benchmarking the performance of their models, and ensuring they maintained the “state-of-the-art”. It also enabled them to empower their team of data analysts to implement AI.

For their logistics project predicting parcel delivery delays, he noted that,  “even though our data scientists had begun building a model in-house using manually coded Python, we ended up using DataRobot”. The whole process ended up taking 10 weeks from start to finish, ultimately resulting in a higher NPS and cost reductions for our call center. DataRobot also played a key role in the release of new strategic customer services. Using AI, they permit some consumers to pay for their purchases in tranches.

Arnaud spoke passionately about the value that had been delivered and the success achieved.  DataRobot allowed his team to become “a lot more agile in the way they prepare, build and deploy models”. The business users also valued “avoiding the black box effect and producing useful prediction explanations”. Overall, the platform delivered on its promises on “model performance, ease of use, and ability to put models into production”. DataRobot beat all competing approaches they tested, often producing more accurate predictions, faster.


In Madrid, Rafael Gonzalez-Iglesias, Head of AI at Fintonic shared their experiences of using the DataRobot AI Platform. Fintonic is one of Spain’s leading Fintechs with over 500,000 customers aggregating their bank accounts on their platform. They’ve raised $50m to date, completing a Series C led by ING earlier this year.

Rafael took us through the lifecycle of an AI project, explaining how DataRobot improved control. “DataRobot helps you avoid lots of boring procedures by simply automating the work” which helped Fintonic to democratise Data Science (involving a wider range of people in projects) and amplify the productivity of the Data Science team - “You can easily see quick results and test them”.

Rafael concluded this meant a “lot of time is saved” in delivering AI projects and “you can increase productivity in an enormous way”.