At the Everyday AI New York conference, when asked for one word from panel moderator Christina Hsiao (Product Marketing Director, Dataiku) to use as a bridge between the analytical present and the Generative AI future, the three panelists responded with “pivot,” “transformation,” and “reimagine.”
Linda Lillard (Managing Director & CDO at BNY Mellon Investment Management), Jody Porrazzo (Director of Statistics & Data Science at Consumer Reports), and Shah Nawaz (CTO & VP, Digital Transformation at Regeneron) shared powerful insights on how they embody the foundations for Everyday AI — democratization (widespread access to analytics and AI), acceleration (of AI project creation and management), and trust (in AI projects and programs). Go further for the full video and a recap of some highlights from the session.
Democratization in Practice
Dataiku enables organizations to drive democratization via tooling that supports multiple personas (from code-first experts to low-/no-code business users) and plenty of built-in helpers and AutoML that make advanced tasks like machine learning, Generative AI, and application development accessible to all.
BNY Mellon Investment Management is set up with a unique, multi-affiliate structure that is efficient for investment management but presents greater challenges for data acquisition and distribution. Challenges include silos (the team is unable to leverage data available in other central data warehouses without copying data), persistent data gaps (the team still relies heavily on spreadsheets, emails, and manual processes to bridge data gaps), and difficulty with standardization due to similar data in different formats from IMFs, asset servicing, and vendors.
Dataiku is a very important, if not the backbone, of this data mesh infrastructure. Dataiku is integrated with our data governance tool and data fabric tools to create a very cohesive environment that allows us to virtualize, stitch, curate, and provision the data to the end consumers. Moreso, the environment allows us to use the same platform to give our end users the ability to search, discover, request, and get access to the data. So the environment is a win-win for the business users and technology. You basically have Dataops, Devops, and MLOps working together in the same environment.”
-Linda Lillard, BNY Mellon Investment Management
The team at BNY Mellon has created a self-service, governed data platform with over 300 users, the majority of which are not trained engineers but rather tech-savvy business users who want access to data in their domain for advanced analytics, BI, and AI solutions.
Acceleration in Practice
Overcoming speed-to-value hurdles is a massive day-to-day challenge for organizations on the path to scaling analytics and AI initiatives. Dataiku enables organizations to accelerate via Dataiku Cloud (no complicated configuration, no infrastructure setup), reusable components, integrations, & solutions, and a single platform — meaning no pain stitching together disparate tools.
Consumer Reports (known for independent product testing and consumer-oriented research) embodies acceleration in practice with their initiative to integrate data science and AI into the full range of core activities by 2026. They plan to focus on prescribing what products and product clusters to test and when, what content to write, when to release that content, and reporting and analyzing the impact the content has on their revenue streams. In the six months since implementing Dataiku, they have:
- Streamlined nine disparate data sources into one unified data view
- Moved from weeks to hours to identify high-opportunity products and product clusters
- Gone from one month to train models to hours on containerized infrastructure
- Consolidated five tools for data prep and dashboarding to one platform: Dataiku
Trust in Practice
To build trust in AI projects and programs, Dataiku enables organizations via teamwork and transparency in a single, centralized workspace, explainability and safety guardrails, and robust AI Governance and oversight. At Regeneron, the complete picture of information includes different voices: patient, clinician, and scientist data. From genomic and demographic data to clinical data and real-time biometric data, Regeneron is focused on connecting billions of data points to accelerate and improve confidence in critical decisions.
Thanks to technology frameworks, people, and processes already in place from the hundreds of successful ML and deep learning-powered solutions Regeneron already has in production, they are able to quickly capitalize on the opportunities afforded by Generative AI and LLMs, rapidly deploying applications for use cases such as document summarization, medical assistance, commercial ad and concept testing, competitive intelligence, multimodal analytics, and more.
Where does trust come into play for Regeneron? They use governance and adoption as key levers for trust, ensuring explainability, drift and bias monitoring, observability, and more. On the Responsible AI side, they leverage AI Governance, acceptable use guidelines, model registry, technical standards, and more.
Anything that we’re developing, designing, deploying, we take that very seriously. Any model in production or that is going to get deployed, we ought to be able to explain how it actually works. Before any model actually goes into production, we ought to have all of the process around it to manage its bias and its drift. And on the change management side of the house it’s an ongoing dialogue, from training to education to seminars.”
-Shah Nawaz, Regeneron
Your Turn to Move From Theory to Practice
The practical examples from BNY Mellon Investment Management, Consumer Reports, and Regeneron emphasize how these organizations are driving the foundations for Everyday AI via democratization, acceleration, and trust. Further, they leverage Dataiku to break down data barriers, streamline and accelerate processes, and build trust in their AI initiatives. As we continue to propel forward in this age of Generative AI, it’s clear that these principles will continue to play a crucial role in shaping the future of analytics and AI.