Building narrow AI models for specific uses: HCLTech

In an interview, Kalyan Kumar, global chief technology officer and head of ecosystems at the software services provider, underlined how businesses are finding early use cases for generative AI, what the metaverse hype cycle has transformed into, and why consumer technology markets are ahead of the enterprise tech adoption curve. Edited excerpts:

Given the slowdown in tech spending across India’s IT services industry, what are the key tech adoption trends that HCL Technologies is seeing among clients?

The evolution of cloud deployment models continues to happen. We’re also looking at an explosion of 5G and edge technologies. We also think connectivity at scale is the other big trend. Quality of access to infrastructure and networks is becoming a fundamental piece. Limitless experiences are also going beyond the metaverse—the latter is something that got overhyped in between, but it’s about the experience and how we want to consume experiences across AR, VR, MR and XR. Cybersecurity is becoming more relevant and a part of our fabric.

Then, there is the power of AI. The way I look at it, AI is tied to the cloud. The latter creates the fabric for data, and you need data for AI. Within AI, a small subset that’s making a lot of noise is generative AI, which is taking the focus away from a lot of other core things around AI.

From a technology perspective, there’s quantum computing that holds the potential to transform a number of tech trends that we see today. Green tech and sustainability continue to be very important. Last but not the least is the buzz around space-tech.

At a larger level, we’re seeing AI’s journey from silicon to software. Clients are reimagining AI holistically and looking at significant adoption of low-code/no-code platforms. Sustainability is becoming a big theme, too.

What investments is HCL Technologies making in building generative AI capabilities?

At a service provider level, for HCL Technologies, the focus is to co-create, infuse and embed AI and generative AI capabilities—by working with customers from a silicon stage to processes. Co-creation involves building technologies for our customers by leveraging our engineering capabilities. Then, we’re helping them apply these techniques to the infrastructure that they have through operations, cloud platforms, cyber security, applications, data-driven processes and business processes.

We’re doing this by infusing tech that’s being built by the hyperscalers (such as Microsoft, Google and Meta)—if you see the public generative AI capability today, that only exists majorly with the three big hyperscalers. But for that, companies will still need access to Nvidia chips. Today, the hottest demand in the market is getting access to chips.

With HCL software, we’re creating some narrow AI models—and not using large language models (LLMs). We’re doing this to be very focused on products and applications. Then, we’ll embed and plug in public generative AI technologies—such as a marketing stack, an example of which is the HCL Marketing Cloud powered by our Unica enterprise automation platform.

We are creating selectively niche capabilities and domain-specific AI and infusing that into our products. Customers are also looking for rapid ideation to prototype processes, for which we have our GenAI labs that allow clients to quickly try out and prototype applications. This is because customers are still figuring out ways to apply the technology.

Are companies still at a preliminary understanding of generative AI?

Enterprise technology adoption was the benchmark in the first 10 years of the 21st century. I think enterprise technology is a laggard. It is struggling and huffing, and it’s the consumer tech that’s led from the forefront.

Even if you look at generative AI, the hype really started with Microsoft trying to use it to make Bing relevant and with OpenAI’s ChatGPT. The maximum usage of this technology has so far come from consumer deployments. That’s where the real value is, and there’s a lot of economic value addition from the consumer side.

In the consumer world, the consumer is the product, and the data that they generate feeds the public data set, which collectively helps everyone. In the enterprise segment, it works differently—you have to apply it atop the enterprise’s data. A lot of pilots and proofs-of-concept are happening among companies that need generative AI at scale, and HCLTech is doing deep work with them—all the way from creating custom silicon up to the overall AI stack. But in the enterprise, customers are realizing that data engineering is becoming a very important discipline to make sure that generative AI is adopted at scale. If you don’t, you run into the risk of exposing your data to the public realm.