Wrap Up | Slalom Triangle Talks | Data: The Fuel for AI Engines

Last week, Slalom Raleigh hosted our 3rd Triangle Talks speaker series in our office. With Generative AI (GenAI) continuing to be the most transformational topic on the minds of our clients, our latest session focused on how enterprises require a strong data foundation to realize the promised benefits. Many of our client discussions on the topic of GenAI quickly turn into a conversation about the underlying components of a strong, modern data platform to enable it. Our goal of this session was to bring insight into the components of a modern data platform and to hear directly from three Raleigh business leaders about how they are approaching GenAI with data at the forefront.

Easy to Experiment, Not as Easy to Scale

As an individual, it’s easy to navigate to ChatGPT on the web or any other public LLM (large language model) and experiment with GenAI. The data behind the LLM has already been curated, cleansed, and segmented, the model has been trained and fine-tuned, and the results are fed into a continuous feedback loop to enhance the next version. But when you want to apply the same possibilities to enterprise use cases, there are a lot of factors to consider. The companies that scale GenAI proof of concepts into production at pace for internal and customer-facing applications will be the ones who ultimately realize the most value from GenAI.? What separates the “fast” companies from those that struggle? Modern Data Platforms.

Clients invested in modernizing their data platforms and processes will be in the best position to capture the value from GenAI. At a high level, a modern data platform is a cloud-based system that has the ability to Ingest, Store, Process, and Publish data for different internal user personas as well as customer-facing applications. ?It incorporates the benefits of the cloud, like scalability, elasticity, and resiliency. The components of a modern data platform include aspects such as Data Quality, Data Governance, Master Data Management, Integration, Architecture, Security, and BI/Analytics. These components are also central to GenAI at scale.

When we start to go deeper on scaling GenAI pilots, some of the typical questions are:

• What Large Language Models (LLM) are to be used, and how are they accessed?
• What training data will be used?
• Where will data reside, and what data is shared?
• What security measures and policies are in place to protect private data?
• How can we measure bias and ensure appropriate responsible AI?
• How are GenAI models integrated into internal and customer-facing applications?
• How are existing talent/skills being updated in an organization?
• What is the operating model at scale for GenAI?

By having a modern platform in place, the answers to these questions will be easier to resolve across the Business and IT ecosystem. Typically, we find clients identifying gaps in their core data platforms as they go from experimenting with GenAI to scaled production. Therefore, we commonly come back to helping our clients mature their data platforms and processes in parallel to GenAI experimentation.

Client Perspectives

We were joined by three different executives in the Raleigh community to share their thoughts on modernizing their data platforms to power AI. Some of the themes that resonated with me from our interactive panel included:

• Many enterprises are still trying to modernize their data platforms and not at the level of maturity they aspire to be at. There is still much work to move legacy data to the cloud and make it available to different users. The good news is there is more support to address the gaps, in large part due to the growing demand for AI products.
• The human factor. Enterprises must invest in their people to future-proof their workforce. Content for both technical and business audiences on data and AI should be integrated into career progression. In addition, applied AI/GenAI is not going to be successful without change management programs to bring people on the journey.
• Agile collaboration unlocks hidden value. Our panelists were clear that interactive, cross-org collaboration to define objectives, data, and outcomes for AI projects yielded better results and commitment to a shared vision.
• Executive commitment is needed at the highest levels for success in AI. Senior leaders need to support and measure progress against defined goals and metrics.
• Responsible AI. It is the responsibility of every organization experimenting with AI to ensure bias is removed from the system. AI needs to maintain a human element in the process/creation/application so that it’s used for the good of all.

GenAI and Modern Data Platforms work together with each other, and GenAI will reinforce the need for enterprises to rethink legacy ways of working, managing data, and making new capabilities available to customers. Data is the fuel for GenAI. Data Platforms deliver the fuel.

Brian Jakubowski