Enterprises Can Accelerate AI Adoption Without Full System Modernization

iconLongbridge
Share
AI summary iconSummary
Enterprises can boost AI adoption without full system overhauls, according to Singtel’s Vinod Joseph Cherunni. Speaking at the Singapore Business Review Technology Enterprise Summit, he urged companies to apply AI now, not later. He pointed to data fragmentation and the need for unified definitions and knowledge graphs. His "model garden" strategy blends open-source and proprietary models to cut costs and avoid vendor lock-in. Blockchain adoption and AI integration can move forward in parallel, he said.

A mix of frontier, open-source, and open-weight models can power enterprise systems.

Enterprises can push ahead with artificial intelligence (AI) across a hybrid mix of legacy and modern systems, rather than waiting for them to be fully modernised, according to Vinod Joseph Cherunni, group vice president for AI/data, cloud and enterprise architecture at Singtel.

Speaking at the Singapore Business Review Technology Enterprise Summit on 12 November, Vinod advised attendees not to wait for systems to be completely AI-ready to start their journey.

“The first thing is [to] reimagine your process, change your workflows, and use AI there." he said. "Otherwise, you're never going to get output as well."

Vinod said enterprise data remains siloed and fragmented across different applications, making it difficult for AI agents to work across business functions.

To address this, he said data “needs context,” meaning consistent definitions across the enterprise, known as ontology, as well as knowledge graphs that capture the relationships between business elements such as customers, products, and systems.

Enterprises may blame AI models for poor outputs and hallucinations when the underlying problem is inaccurate data and a lack of context, Vinod said.

“This is exactly why a lot of [proofs of concept] and pilots fail,” he said.

“Every time we have the best data, the best, cleanest possible data with the context, the number of tokens that you burn [is] extremely less,” he added.

Vinod described a “model garden” approach that combines open-source, open-weight, and frontier models.

Enterprises can develop a layer that contextualises prompts sent by agents and applications and forwards them to different models based on different characteristics.

Spreading workloads across models would not only prevent vendor lock-in but also manage cost and sovereignty, he said.

Customer data remains local, with on-premises models that can operate within a sovereign infrastructure, whilst a responsible AI framework ensures that the models do not receive customers’ specific personal information.

Vinod told attendees to move ahead with autonomous operations across their IT environments, with the aim of remediating problems as quickly as possible and identifying security incidents before they occur.

Even a fair reduction in the number of incidents can be a significant advantage, he noted.

Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information. Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.