Ubisoft puts players at the center of everything it builds. Every interaction across franchises like Assassin’s Creed, Just Dance and Rainbow Six generates data that supports recommendation models, gameplay analysis and cheat detection, all in service of better player experiences.
As demand grew across its employees and multiple simultaneous productions, Ubisoft’s aging Hadoop ecosystem and fragmented custom tooling could not keep pace. With the Databricks Platform, Ubisoft replaced a patchwork of solutions with one governed flow connecting data pipelines, analytics and applications. Teams can now self-serve trusted data, reduce time to insight from days to seconds and turn data into the strategic asset the company needed it to be.
Game data was outpacing the people who needed it
Ubisoft's mission is to deliver meaningful, evolving player experiences across some of the most recognized franchises in gaming. Achieving that at scale requires treating data as a strategic asset, accessible to every team that shapes what players see on screen. For a company of 17,000 people, the challenge was never generating data. It was getting it to the right people quickly enough for them to act on it.
A legacy platform stretched to its limits
Ubisoft's legacy Hadoop environment, built from open source libraries and custom tooling, served the company well but became harder to scale as multiple simultaneous productions grew in computational demand. Insights passed through disconnected processing, transport and exposure modules before reaching teams, and business data flowed through a custom ingestion tool managing more than 4,000 pipelines where deliveries could take 30 minutes to a few hours.
"The demand kept accelerating and the system couldn't keep pace. Every patch we made bought us time, but it didn't solve the underlying problem," said Tsvetan Petkov, Technical Director of the Data Platform at Ubisoft.
Data access that depended on who knew SQL
Level designers and gameplay programmers depended on analysts to query the platform and return results, a workflow that was standard but hard to scale. After a playtest, a single analyst might field hundreds of similar requests, and answers could take days to come back.
"The data was always there. The barrier was that you needed SQL to reach it, and most of the people who needed answers didn't have that skill," said Thomas Donadei, Technical Product Manager at Ubisoft.
One platform connecting pipelines, insight and applications
Ubisoft needed a platform that could handle scale, bridge the gap between data production and consumption, and open access beyond those who knew SQL. The team moved to the Databricks Platform with Unity Catalog as the governed foundation, giving every team a single path to trusted data and a single set of access controls, rather than the custom-built layer they had outgrown.
Migrating 4,000+ pipelines off legacy tooling with Lakeflow
The team adopted Lakeflow, a managed ingestion and orchestration service, to start untangling the pipeline layer. More than 4,000 business data pipelines ran on legacy tooling, and Lakeflow is giving the team a path to migrate and standardize them without rebuilding each one from scratch. Instead of writing custom code for every new data source, teams configure connectors and transformations visually, and new sources reach the business in minutes.
"Four weeks after the first demo, people who had never built a pipeline were managing their own data ingestion," said Tsvetan. "That kind of ramp-up simply wasn't possible before."
Lakehouse data is available to apps in seconds
To get that data to applications just as quickly, Ubisoft paired Databricks Apps with Lakebase, a serverless PostgreSQL backend that automatically syncs data from the lake. Changes processed upstream are available for consumption in as little as two seconds, eliminating the export pipelines the team previously maintained. For smaller teams without dedicated DevOps support, that changes what's possible. They can stand up a data-driven application, connect it to Lakebase and scale on demand without managing infrastructure.
"Teams that would have spent weeks provisioning infrastructure were up and running in hours. Deploy, connect, push data. That was it," said Tsvetan.
Natural-language access for non-technical roles
Even with cleaner pipelines and faster application backends, designers and gameplay developers still needed SQL to query the data. To close that gap, Ubisoft piloted Genie on a production game, giving those roles a way to ask natural-language questions directly after playtests, rather than routing repeat requests through analysts. The team is now expanding Genie across additional productions and datasets.
"The goal was always to make data accessible to all 17,000 employees," Thomas said. "Genie is the first tool that actually makes that realistic."
Faster answers for the people closest to the games
In one pilot, the data director estimated that roughly 60% of analysts’ time previously spent fielding repeat playtest questions has been reclaimed. Designers and developers who once waited days now get answers in seconds, and the analysts who used to handle those requests have moved on to deeper work. The platform team experienced the same shift firsthand when they used Genie Code to build a dashboard tracking adoption and costs for the company's AI Gateway, a project that previously would have required a dedicated analyst.
"The questions haven't gone away," Thomas said. "There are just as many as before. The difference is Genie answers them, and our analysts can focus on the work that really moves the needle."
Speed from ingestion to application
Lakeflow and Lakebase extend that same speed across the rest of the chain. New data sources that required custom development now reach the business in minutes through Lakeflow, and syncs to Lakebase for serving into applications in as little as two seconds, down from 30 minutes to hours. When the company needed to give its development workforce governed access to cloud-hosted AI models, the platform team used that same foundation to stand up an AI Gateway and move from planning to production in two to three weeks.
The road to governed agentic workloads
With the core platform in place, Ubisoft is turning its attention to what comes next. Lakeflow will become one of the company's main orchestration tools, Lakebase the primary backend for new Databricks Apps, Genie a path to broader self-service analytics, and Genie Code a way for technical teams to accelerate governed data and AI development. "The industry was moving fast, and we estimated we were three years from where we needed to be," said Tsvetan. "That gap is virtually gone. Our last mile is governed by agentic workloads, and with Databricks, we'll get there."