As an AI and data expert who has worked with some of the world's largest enterprises, I've seen firsthand the challenges that organizations face in scaling their AI capabilities.
The key to success is having the right data foundation in place. Put simply, there is no AI without data. But data alone isn't enough; organizations need to bridge the silos between data, AI and infrastructure to get the most value out of "traditional AI".
Most organizations, especially non-digital natives, are missing the connection between these three efforts, which is hindering scaled innovation ambitions.
Bridging the gap
In today's digital world, organizations are focusing on monetizing data as an asset. To achieve this, data and AI strategies and teams have been tightly linked to deliver business outcomes. There needs to be a deliberate business strategy on how to leverage the technology innovations happening in the world of data, AI and infrastructure.
The basic value chain of data and AI is simple: Data is converted into insights, which is converted into actions and decisions to create solutions and services.
For example, with traditional AI use cases like personalization, an AI system provides insights based on someone's personal data, like from social media, and then generative AI (GenAI) models produce the relevant content for the specific individual.
What's often missing from this unified data and AI strategy is infrastructure. Both data and AI are infrastructure-hungry, and while storage costs have become lower over time, if organizations don't consider the infrastructure strategy, then the ROI becomes highly questionable.
To put this into context, we helped a hospitality client create a digital assistant. We managed to build it out in a matter of days. However, since the awareness about infrastructure costs was low for this client, it ended up costing almost $300 an hour to run it.
Very few organizations have an enterprise-wide unified data strategy that connects both AI and infrastructure, as well as structured and unstructured data. To help organizations address this challenge, HCLTech has developed the Enterprise AI Foundry, which aims to execute AI initiatives with a clear view of costs and requirements by helping enterprises bring together data, AI and infrastructure in a cohesive way.
Organizational models also need to pivot to break these silos. Agile DevOps ways of working and embracing change management are imperative to scale AI to transform your business.
Connecting the dots
Once a business strategy is created, to build the connections between data, AI and infrastructure, there are four steps that organizations can follow:
1. First, understand that you need to shift from collecting to connecting the dots. A mindset shift needs to occur.
2. Identify the big problem you're trying to solve and the use cases that will create the maximum business impact, rather than just collecting data and building AI models and use cases in isolation.
For example, a bank we worked with wanted to run AI models to enhance trade surveillance. This required us to look at the problem statement holistically, starting from the desired output and working back from there.
3. Determine the optimal way to run AI models — either by sending the data to the model, such as with OpenAI's ChatGPT, or getting the AI or the model to the data, which resides in a private, secure environment. Either route depends on factors like data sensitivity and cost, and infrastructure must be a consideration, as AI models consume a lot of GPUs.
4. Strategically ensure the AI, data and infrastructure teams are aligned and working together to solve the problem rather than operating in silos. Otherwise, organizations will end up with multiple flavors of execution from a technology standpoint for the same project.
Scaling AI and innovation
Once the dots are connected, organizations can start scaling AI and innovation initiatives throughout their enterprise by following these steps:
1. Shift from a use-case-driven approach to a capability-driven approach, building reusable AI capabilities like conversational AI and voice analytics for internal or external service desks. A company looking at a 100 use cases can group them together by creating 10 distinct capabilities.
2. Establish a centralized data, AI and infrastructure team to build the core foundation, platform and capabilities, while allowing business units to build their own AI-powered applications on top. This is crucial for establishing efficiency and consistency.
3. Ensure the technology approach is aligned to the organizational model.
4. Democratize the use of AI across the organization, making it easy for non-technical employees to leverage the capabilities and create value. It's important to understand how to make it easy for the employees to consume AI create value.
5. Focus first on targeting the low-hanging fruit, which is driving efficiencies, before transforming employee and customer experiences and then creating new products and services.
The shift has begun
The organizations that will succeed in scaling their innovation and AI ambitions need to shift from simply collecting data to connecting the dots between data, AI and infrastructure through a simplified, outcome-focused approach.
By adopting a unified data, AI and infrastructure strategy, organizations can lay the foundation to effectively scale their AI ambitions and drive tangible and sustainable business value.
A shift in attitude is taking place and organizations are realizing they need to be much more strategic about how they leverage these three areas to scale innovation effectively and achieve the business' goals.
It's advisable to start with the outcomes and then work backwards to figuring out what datasets are needed and how to industrialize that data with AI.
The good news today is that cloud, AI and data management technology is now available to change ways of working, bridge connections and drive innovation at scale. A decade ago, the tech was not available, and big data projects failed to deliver the desired ROI.
