You picked up a Rs 80 crore Indian bank deal recently. How is the business traction currently for Intellect Design in the bank category in India and among public sector banks (PSBs)?
It's quite positive. Indian banks have been very positive on picking up the high-quality business impact technologies. Five years back, the trend was different, as they were going with a cost-conscious approach. But even now, some of the PSBs are unable to change their mindset in terms of business impact measurement. In fact, they do not have business impact measurement. So, something is wrong in the public sector units, which the Department of Finance has to solve.
It's interesting that you mention this, as the sector appears to be focused on non-performing asset (NPA) improvements and celebrating it, while simultaneously, traditional banks seem to be misguided in their prioritisation of technology…
It is a huge issue, one that often leaves me feeling disheartened. We recently encountered a situation where, without naming the bank, our quoted price was significantly higher than a competing vendor's-by 40 per cent, to be precise. The bank felt compelled to opt for the lower-priced offer. When we confronted the bank's CEO about the rationality of such a decision given the quality of technology at stake, the response was a resounding acknowledgment of the mismatch between price and value.
This begs the question: Why engage in a contract with a vendor who operates at a loss? Sure, you can bind them contractually, but can you expect a vendor bleeding money to provide sustained support? The impact goes beyond contractual obligations; it affects the quality of technology delivered. What's often overlooked is that the nominal amount banks allocate for technology-just 1 per cent of their business impact-can have far-reaching consequences if invested in the wrong solutions. If they choose the wrong technology, they are gone for the next 10 years. So, private sector banks will kill public sector banks. It is a serious issue. I do not know in which forum it can be addressed, but it's difficult for them to see this issue.
Where does the issue lie?
I find that bank managements' hands are tied. For Intellect, it does not matter too much because our dependency on PSBs is limited. If the banks looked at business impact, then they would give their specifications much more in detail. But they give very gentle specifications, which is not a good indication.
Celebrating NPAs is alright for now, but without technology NPAs will grow. Public sector NPAs happen in a five-year cycle and hence, banks must heed to the need of the hour.
Coming to Intellect Design, how has the company leveraged generative AI technology?
AI is often misunderstood in banking context. It boils down to trusted data and expert agents network supported by technology (network is supported by technologies, including reasoning and providing an audit trail for recommendations). Our AI suite extracts reliable data from documents, aiding expert agents in making informed decisions based on bank policies. While generative AI is invaluable for creating general systems and managing knowledge, banking operations, which involve tangible financial transactions, require a balanced approach incorporating both generative AI and machine learning algorithms. Integrating these elements allows us to develop a comprehensive AI suite tailored to address specific challenges within the banking sector.
The technology spotlight has been on generative AI since 2022-end. But considering your eMACH.ai platform launch happened in February 2023, can you clarify if generative AI was part of your plan, and was added initially?
Generative AI wasn't initially part of our plan. We were focused on developing T1 to T4 level technologies first. Generative AI came into the picture in 2022 and we swiftly integrated it wherever applicable alongside our existing technologies. This approach allowed us to address problems more efficiently, particularly in generating recommendations from T5 to T8 level of technologies.
What models have you leveraged in your platforms?
We have used some open LLM models. We also use OpenAI's and Anthropic's LLMs to solve problems.
