Barbara Latulippe, Chief Data Officer at Takeda Pharmaceutical Company Limited, speaks with Tracy Gusher, AI, Data and Automation Leader at EY, in a video interview about her role, what it means to have AI-ready data, the three pillars that drive AI-readiness, and the association of data with responsible AI.

Takeda is a leading values-based, R&D-driven global biopharmaceutical company.

Latulippe shares that her role as CDO is exciting as she leads platform-as-a-service, data-management-as-a-service, AI, and advanced-generative-AI-as-a-service. The role encompasses everything data — from ingestion to data markets and data products. The advent of generative AI (GenAI) has added new dimensions to the role.

When asked what it means to have AI-ready data, Latulippe states that AI cannot exist without data, and it all begins with a strong data foundation. To be GenAI-ready, it is critical to have robust data governance, and Takeda has expanded that focus to include analytics governance, she adds.

In addition, data quality is paramount, and the right data should be available for the right purpose. This involves maintaining contextual and high-quality data and implementing processes like data certification, whether registered, validated, or certified. This supports self-service for business partners, enabling them to access the right data at the right time via the organization’s data marketplace.

According to Latulippe, being AI-ready is excelling at the basics of data management. These include understanding data sources and data acquisition, and optimizing purchased and incoming data, especially with the growing unstructured data.

Elaborating, she says that Takeda has established a new department under her team focused on data acquisition. Prioritizing data quality, she says that not all data hold the same value. Adding further, Latulippe says that Takeda’s data quality is aligned by domain, with 21 domains in total, and the strategy is driven by three strategic pillars.

1. Accountability — ensuring good data quality, defining KPIs, maintaining contextual data, certifying it, and controlling access, she notes.

2. At scale — which translates to being agile and assessing whether the platform and governance practices can scale. The company introduced new requirements in this pillar, especially around legal, ethical, and compliance considerations for responsible AI.

3. GenAI for all — built on the foundation of excellent data management practices. With domain accountability, business leaders are assigned to partner closely with data, digital, and technology teams. This collaboration ensures the business voice drives domain usage.

Out of the 21 domains, five to six are mastered through master data management (MDM) environments, while the rest are managed through data quality rules, says Latulippe. Takeda is also developing an extensive data stewardship matrix, working across regions, local levels, and digital teams to identify the data needed for effective governance and relevant GenAI use cases.

Shedding light on what Takeda does, Latulippe states that the organization is committed to being the leading science-driven, digital, and data-focused pharmaceutical company. It places patients at the core of everything it does, which is why the domain strategy is centered around healthcare providers, patients, suppliers, products, and brands, ensuring engagement in a regulated environment.

Takeda’s goal is to deliver lifesaving therapies to patients more quickly and target them more effectively during clinical trials. One of the organization’s greatest challenges is identifying and enrolling the right patients for these trials and monitoring them throughout the patient pathway. Accurate data, combined with GenAI, will accelerate this process, enabling the company to reach patients sooner and provide them with the critical care they need.

Emphasizing the association of data with responsible AI, Latulippe first shares about achieving significant success with the responsible AI framework, which is widely accessible across Takeda. The approach includes a robust data governance structure led by the Global Data Council which oversees all aspects of data.

Complementing this, Takeda established the AI and GenAI Steering Committee, comprising representatives from across the organization. The committee acts as the governing body for AI and GenAI initiatives, where executive-level decisions are made to ensure the responsible implementation of AI and GenAI.

Latulippe mentions treating GenAI as an extension of traditional AI and machine learning rather than separating them. Expanding the governance has been key as it brought in many new dimensions, such as ethical use and ensuring unbiased datasets. She also mentions developing a data marketplace that serves as the central repository for GenAI and analytics models and includes a vector marketplace, an API marketplace, and expanding capabilities for agents and prompts.

Integrating traditional data and GenAI governance enhances the data lineage, having the right metadata, and then monitoring these through the responsible AI framework. She mentions an ongoing ML and LLMOps program for continuous monitoring and requires sign-offs from relevant groups to ensure algorithms are responsible.

Many current algorithms focus on improving internal productivity, and Takeda has developed a classification system to determine the appropriate level of governance required before models enter production.

Further, Latulippe mentions launching an internal collaboration platform called Takeda.ai, which features an agent and prompt library, learning pathways, and direct access to the responsible AI framework. Moreover, a complementary platform, Takeda.data, is being developed for the data community, she adds.

Adding on, Latulippe states that as companies focus on optimizing AI, building a solid data foundation is critical, along with establishing AI governance early in the process. She states that this approach has significantly impacted Takeda’s efforts and enabled collaboration across the company. Elevating both data and AI fluency has been a critical aspect of this journey.

In conclusion, Latulippe shares that the company initially started with a user group of 1,000 people and is now rolling it out company-wide. This effort also highlights its GenAI use cases, fostering self-service discovery.

CDO Magazine appreciates Barbara Latulippe for sharing her insights with our global community.