Aditya Birla Capital (ABCL) is combining internally developed artificial intelligence (AI) tools with solutions from external partners to improve business operations, while avoiding dependence on any single technology platform, said Pankaj Gadgil, Head of Digital Platforms and Payments Strategy at ABCL and MD & CEO of Aditya Birla Housing Finance Limited.
Combining in-house development with external solutions
Gadgil said companies should balance building their own tools with adopting solutions already available in the market. ABCL has developed a platform for sales and service assistance, supported by separate databases for its businesses and multiple large language models (LLMs).
The company has also developed speech-to-text and other voice tools while leveraging solutions from technology partners. Its internally developed tools include WorkIO, which manages internal operations; TraceIO, which tracks customer journeys across digital platforms; and FinEngage, which enables employees to trigger automated customer outreach at specific stages of loan or insurance processes.
“It always has to be a hybrid between build and buy,” Gadgil said, emphasising the need to remain open to solutions available in the market.
Keeping technology platforms flexible
Gadgil highlighted the importance of maintaining open technology architecture as AI models and platforms continue to evolve. ABCL uses both open-source and closed models and is exploring the role of smaller language models (SLMs) in handling operational tasks that may not require larger models.
“I think a lot will happen in the area of SLMs replacing LLMs, because some of the operational problems we are trying to solve with LLMs do not require them. SLMs can handle them,” he said.
He added that AI tools must integrate with existing systems, including Salesforce and loan platforms such as Lentra and FinBox, through application programming interfaces (APIs) that enable them to access and update information.
Data and governance remain critical
Gadgil said ABCL is working towards an enterprise-wide “knowledge IQ layer” to connect data with its meaning and relationships across formats, including emails, PDFs and images.
“We increasingly realise that to get greater impact from AI, most of the work that now needs to be done, I think, is on data,” he said.
He added that scaling AI effectively also requires employee adoption and robust governance covering tool access, accuracy, costs and incident management.
