Bring the Large Language Model to Data, Not Vice Versa

Enterprises should not consider embracing GenAI or LLM just because it is the coolest and the most hyped technology, but because it solves a business problem and creates opportunities 

opening
info_icon

In the ever-evolving landscape of technology, the allure of the latest and most hyped solutions can be tempting for enterprises. However, the adoption of cutting-edge technologies, such as generative artificial intelligence (GenAI) and large language model (LLM), should not be driven by trends but by the fundamental goal of solving business problems.

Consider a manufacturing company which aims to reduce machinery failure rates and enhance production efficiency. The information critical to achieving this objective might be scattered across sources, including unstructured documents, like physical reports. LLM can play a pivotal role in extracting structured content from unstructured documents in such a situation, thus offering a tangible solution to a specific business problem. ChatGPT, for instance, follows an LLM model and has proved its worth in varied scenarios.

Snowflake, a data cloud company, sees GenAI as a disruptive force with transformative business potential. Sanjay Deshmukh, senior regional vice president for Snowflake for ASEAN and India, believes that GenAI is the most disruptive technology of the present times whose use cases go beyond the obvious. He says, “One should not judge it by what is happening in the consumer world, since it has the power to help businesses accelerate their digital transformation in many ways.”

Snowflake recently acquired Applica, an AI-based platform for document automation, and built it further as Document AI. Deshmukh says, “A business manager can give Document AI machine service reports for the past year and ask it to derive insights about the failure rate of parts. This analysis can improve production efficiency of the plant by enabling the manager to take timely calls on fixing a particular part, the entire machine or changing the supplier.”

Beyond the Hype

While businesses are beginning to understand the importance of LLM for their sectors, they have developed the same angst about it as they have for adopting any cloud-based solution. This is specifically true of enterprises which handle sensitive data in large buckets. Their concerns include potential risks associated with sending business data to external servers, uncertainty about where the data will be processed and stored and a possible lack of control over data insights.

Deshmukh says that as a leading player in the LLM space, his company is aware of this challenge and reassures its clients. “We are educating them, so they understand the foundational premise needed to first build a data strategy to have a successful AI strategy, because data powers AI,” he says.

It becomes essential for businesses to know that large data, in any case, needs specialised analytics for it to make any actionable sense. With LLM, it becomes almost impossible for businesses to have in-house solutions. Snowflake sees an opportunity in assisting organisations worldwide in accelerating their digital transformation by emphasising the importance of data in AI-driven decision-making.

Deshmukh emphasises that the key to using AI lies in its alignment with business imperatives of an organisation, which lies at the core of Document AI, since the model resides within a company’s security perimeter, ensuring that data stays within and insights are derived from proprietary data.

India Holds the Key

Data AI companies know the importance of India in their revenue strategy. Deshmukh feels that India’s burgeoning start-up ecosystem, marked by innovations in online retail, edtech and gaming, presents opportunities for his company to grow and help businesses leverage its solutions for building and scaling their offerings.

“Indian start-ups are not just catering to the local demand. They are building SaaS solutions which impact global markets. We are investing in India to help these companies build a solution and leverage Snowflake on top of that,” he says.

The reverse is true too. If a global corporation is already a client of a data AI company like Snowflake in other markets, it makes sense for them to integrate the profiles of their Indian users with global data and run analytical tools on them. A multinational data AI company offers that advantage to such clients.

Power to Data Scientists

Data-heavy companies are still fighting the legacy issues of managing unstructured data. Even recent data is not available in formats that machines can read and analyse quickly. Though these companies, including start-ups, employ data scientists, their efficiency stays low.

“Most data scientists spend excessive amount of time preparing and structuring data for analysis, which is not their job. We want to improve their productivity, so that they do not have to bother about how or from where to get the data,” says Deshmukh.

For this purpose, Snowflake has launched the Snowpark framework, which makes access to centralised data easy. This way, argues Deshmukh, data scientists can focus on building and training models without the burden of data preparation.

This approach brings an added advantage to start-ups, whose core strength lies in processes and services. Access to structured and centralised datasets reduces time-to-market for their services, which can provide them relief in an era of funding constraints like the present times. Since this approach integrates well with the infrastructure of Big Tech players like Microsoft, AWS and GCP, data stays safe, actionable and product driven at any scale.

Legacy businesses and start-ups in India know that they cannot afford to lose opportunity that analysis of large data creates. Once data analytics companies assure them of security and scaling up around their data, these businesses are likely to dive deeper into the potential of their humongous data, something that cannot be done without the help of GenAI and LLM.

Tags