Prologis: Reaching Operational Excellence With Dataiku

Prologis’ AI journey began with a traditional data warehouse setup and evolved into a sophisticated cloud-based environment that uses Snowflake for data warehousing and Dataiku for AI and machine learning (ML) development. The data and analytics team moved their data stack to the cloud with Snowflake to boost analytics modernization. 

The company’s first AI use case revolved around descriptive analytics, including management and operational analytics. The goal was to clearly communicate what was happening in the company. To do this, the team looked at proper data modeling and delivered value from there. As Prologis continued to evolve, the data and analytics team shifted towards more advanced use cases, including geospatial analysis and predictive modeling.

Overcoming Challenges in AI Deployment With Dataiku

Before Dataiku, data scientists at Prologis would build models on their local machines, often with data extracts, and then hand them off to the data and analytics team for deployment. This process was time-consuming and had some challenges related to system dependencies and packaging. It created inefficiencies, inconsistencies, and delays in deploying models into production. Models thus often became outdated before they could be operationalized, limiting their business impact.

In addition, AI efforts were siloed, with only a small group of skilled data scientists able to build models. This bottleneck prevented wider adoption of AI/ML tools across the company and hindered data-driven decision-making at scale. To keep pace with the rapidly evolving logistics industry, Prologis needed a solution that could centralize AI efforts, integrate with its existing technology stack, and empower non-technical users to engage with AI tools, driving faster and more efficient business decisions.

The Solution: Democratizing AI With Dataiku

To streamline this process, Prologis adopted centralized development platforms like JupyterHub servers, which allowed for seamless integration of data science code into production environments. However, the need for coding skills was still a barrier. This led Prologis to explore AutoML and AI platforms like Dataiku, which democratized AI by enabling analysts to build and deploy models without extensive coding knowledge.

Dataiku provided the flexibility, scalability, and ease of integration needed to streamline the company’s data science operations by:

• Ensuring End-to-End AI Lifecycle Support: Dataiku supported the entire data science process, from data ingestion to model building, exploratory data analysis, and the development of web apps and dashboards. This provided Prologis with a unified workspace where teams could move from raw data to actionable insights more efficiently.
• Seamlessly Integrating With Snowflake: Dataiku’s strong partnership with Snowflake, which is central to Prologis’ tech stack, enabled large-scale data processing and optimization. This combination ensured that Prologis could handle complex use cases while remaining flexible and scalable.
• Democratizing AI: With Dataiku’s no-code and code-friendly environment, Prologis was able to empower a broader range of users — including business analysts — to build and deploy AI models. This helped decentralize AI development and allowed more teams to leverage advanced analytics tools in their day-to-day work.

Analysts can come in, they get seamless integrations with all of our data source systems. They can do EDA, build models, build web apps or dashboards. It's very powerful and it's enabled a lot of our users. Jennifer Garcia Lead AI/ML in Data & Analytics at Prologis

By incorporating Dataiku into their AI strategy, Prologis significantly accelerated AI/ML model deployment and unlocked new opportunities to innovate across business units. By partnering with their in-house operational excellence team — which focuses on process engineering, process adherence, and standard work — Prologis ensures that deployments fit into a business process and project efficacy is measured.

One of the most significant outcomes has been the widespread adoption of AI across different departments. Thanks to Dataiku’s flexibility, 85 active users — beyond just data scientists — now contribute to AI initiatives. This has driven efficiency gains in areas such as revenue management and operational forecasting.

Value Generated With Dataiku

Speed and Agility

Dataiku has significantly reduced the time it takes Prologis to build, deploy, and iterate on AI/ML projects. Since implementing Dataiku, Prologis has put over 60 AI/ML projects into production, a 12x increase from just five prior to adopting the Universal AI Platform. This rapid scaling has enabled faster project delivery, improved decision-making, and enhanced operational efficiency across multiple business lines.

The impact on production has been substantial: Prologis has gone from maintaining just five productionalized data science products to 30 projects and 30 APIs in active use. With Dataiku, Prologis can now iterate on AI projects far more efficiently, significantly shortening development and deployment cycles. This has ensured that their AI solutions remain dynamic and responsive to changing market conditions, embedding AI deeply into daily business operations.

Increased Interoperability

Dataiku’s seamless integration with Snowflake has allowed Prologis to handle large data volumes and complex use cases efficiently. This has enhanced operational efficiency and ensured that Prologis can scale AI/ML initiatives without infrastructure limitations.

At Prologis, we are selective about the technologies we embed deeply into our operations. Dataiku and Snowflake are two foundational pieces of our AI strategy. Jennifer Garcia Lead AI/ML Engineer at Prologis

When it comes to emerging technologies like Generative AI, Prologis focuses on tools that enable easy transitions. This includes modular architecture and interoperability, allowing them to plug and play new components as needed without becoming entrenched in any single technology. This approach ensures flexibility and agility as the AI landscape continues to evolve and ensures that the team maintains their focus on initiatives that will move the needle for the company.

Generative AI Initiatives

With Dataiku, Prologis was able to build and rapidly deploy Generative AI pipelines, which led to their development of custom GPTs in their Enterprise ChatGPT platform. This allowed Prologis to develop and launch 10 Generative AI products in production, a testament to the platform’s agility. Through the Prologis Enterprise ChatGPT platform, over 2,000 users are now leveraging AI to streamline workflows, automate processes, and make data-driven decisions in real-time.

Overall, the value created through Dataiku has gone beyond just speed and scale — it has empowered Prologis to make data-driven decisions at all levels of the business, improving efficiency, reducing costs, and ensuring the company remains agile in a rapidly evolving industry.

In a 2024 Everyday AI San Francisco session, “Sticking the Landing: Making the Leap to AI and Surviving to Tell the Tale,” Luke Slotwinski, VP of data & analytics at Prologis, and Jennifer Garcia, a lead AI/ML engineer on Luke’s team, explained how they are infusing AI into their business operations, prioritizing AI use cases, and achieving organizational alignment — all with support from Dataiku.

Watch Video