3 AI Trends from the Big Data & AI Toronto Conference
Organizations are looking for AI platforms that drive efficiency, scalability, and best practices, trends that were very clear at Big Data & AI Toronto. The DataRobot expo booth at the 2022 conference showcased our AI platform with industry-specific demonstrations including Anti-Money Laundering for Financial Services, Predictive Maintenance for Manufacturing and Sales Forecasting for Retail. Swarms of customers, partners, and industry colleagues dropped by to discuss AI-related opportunities within their organizations and discuss three top AI themes.
Monitoring and Managing AI Projects with Model Observability
Model Observability – the ability to track key health and service metrics for models in production – remains a top priority for AI-enabled organizations. As AI-driven use cases increase, the number of AI models deployed increases as well, leaving resource-strapped data science teams struggling to monitor and maintain this growing repository.
“We have built hundreds of demand forecasting models at the store-level, and now my data scientists are spending valuable time babysitting these models instead of working on new projects,” shared the Director of Analytics of a global retailer. Today, his team is using open-source packages without a standardized AI platform. Knowing this, we walked through a demo of the DataRobot MLOps solution, which can manage the open-source models developed by the retailer and regularly provide metrics such as service health, data drift and changes in accuracy.
Later in the demo, the retailer also expressed the need for an easy diagnosis of performance issues, allowing him to swiftly get to the root cause upon being notified of an issue. We dug into interactive visualizations such as the DataRobot drift drill down plot, where users can investigate the exact feature and time period affected by data drift in a model. The demo sparked an ideal reaction from the retailer, who emphasized that such changes will “completely change” how his team spends their time.
Accelerating Value-Realization with Industry Specific Use Cases
One of the biggest bottlenecks to AI adoption is finding appropriate use cases,1 and business leaders in attendance at the Big Data & AI conference echoed this sentiment. While AI is a powerful and dynamic tool with the potential to deliver tremendous business value, identifying the right business use case remains a challenge for many organizations.
For example, conference attendees from the financial services industry expressed the need in their organizations to improve financial crimes solutions using AI. Detecting credit card transaction fraud and detecting money laundering are both examples of financial crimes, however the two use cases require different frameworks. Organizations need playbooks that outline the framework and the steps required to successfully implement a particular use case.
Created from 10 years of experience working with the world’s most pivotal organizations, DataRobot Solution Accelerators are a library of hundreds of AI use cases captured by data science experts based on real-world implementations. These accelerators are specifically designed to help organizations accelerate from data to results. AI leaders at Big Data & AI Toronto explored use cases specific to their industries for inspiration and guidance.
At the conference, I delivered a workshop on anti-money laundering best practices using AI, sparking discussion and inspiration among anti-money laundering experts. They were surprised by the efficacy of AI in identifying a few suspicious transactions hiding among millions of normal transactions. I demonstrated how this “needle in a haystack” problem can be solved by leveraging automated machine learning to rank potentially suspicious alerts enabling AML agents to prioritize investigation of high risk alerts, effectively reducing the number of costly false positives.
Lower AI Barriers with Deployment Flexibility and Interoperability
Any AI conversation is incomplete without the mention of the power of cloud computing. Enterprises are migrating workloads and relevant technologies to the cloud for superior compute power and streamlined operations. Interoperability of the existing technology stack is a challenge most IT leaders are facing as more and more business-critical workloads are moved to cloud environments.
As each AI use case requires a different framework for model development, deployment methodologies are also use case specific. Some use cases require models to be scored in real-time with very low latency, such as transaction monitoring systems. Other use cases, such as marketing campaigns, need to run on large quantities of data, but latency isn’t particularly an issue. As a multi-cloud platform, DataRobot enables organizations to run on a combination of public clouds, on-premises data centers, or at the edge, depending on the business needs.
At the conference main stage, Ricardo Baltazar, Associate Vice President, Innovation Lab at Canadian Tire Corporation presented the retail giant’s journey and success in scaling AI using DataRobot. Canadian Tire is one of Canada’s most recognized retail chains offering over 1700 retail locations, financial services, and e-commerce capabilities.
Leveraging DataRobot AI Platform, Canadian Tire standardized business critical workflows like AI deployment and active monitoring of production AI. Ricardo emphasized the importance of interoperability of technology stack to scale value across the enterprise.
DataRobot at Canadian Tire has lowered AI barriers with the flexibility to deploy models quickly in any environment, and by integrating with other business tools for standardized and seamless workflows. These alignments enabled Canadian Tire to realize deeper business value and build AI trust across the whole organization.
Learn How to Accelerate Business Results with DataRobot AI Platform
Learn more about the DataRobot AI Platform and the ability to accelerate your experimentation and production timelines. Explore the DataRobot platform today!