Snowflake exploring partnership with govt in generative AI: Sanjay Deshmukh

Governments in many developed economies are already collaborating with Snowflake, to support them in their decision-making, said Deshmukh

Snowflake, a US-based data cloud specialist, is expanding in India to meet the technology needs of businesses and government organisations, said SANJAY DESHMUKH, the company’s senior regional vice president for ASEAN and India. The Nasdaq-listed firm is tapping startups, traditional enterprises and state organisations to provide its services in cloud computing and artificial intelligence (AI), he told Peerzada Abrar in a video interview. Edited excerpts.

What impact do you see generative AI having on businesses?

There is a fundamental shift that we see happening with generative AI and LLM (large learning models.) If I look at the old world or what used to happen before ChatGPT, or prior to generative AI and LLM becoming mainstream, the machine learning models were built by data scientists employed by companies in areas such as retail and banking. They were applying those machine learning models to their data. They were doing everything from prediction and recommendation. What has changed with generative AI and largely LLM is that this work of building the models is now being outsourced to external entities. If you look at the foundation models, OpenAI has built them, and there are many other companies that are building LLMs for specific areas. I'm not saying it eliminates the work of data scientists. But now I don't need to spend energy in building a model, rather my effort should be to bring a model that is relevant to my business and see how I can deploy that.

What are the challenges?

If you are a consumer, you can use platforms like Midjourney (an AI service) and ChatGPT for your AI strategy. But if you are a company or an enterprise, you need to recognise that you have proprietary data — intellectual property and customer data — that you have to support and manage. In order for you to build an AI strategy for them, you need a strong foundation of a data strategy because it is the data that powers AI. As we are working with customers, we are trying to educate them that we will help you enable generative AI and LLMs in their business. But the first step in moving that direction is to actually build a solid data strategy. We help them do that with the Snowflake data cloud platform. As enterprises leverage these LLM models, one of the risks is that your data will leave your security perimeter and it may get compromised. Also, these foundational approach models are very broad in nature, from writing Shakespeare's poetry to answering questions about world hunger and climate crisis. For an enterprise trying to solve a business problem, it needs something very specific.

What are the innovations that you have built to address such risks?

Our philosophy of using generative AI and LLM is that the data should not travel to the model. The model should sit next to the data. If the model is sitting next to the data, [then] the data is not travelling outside of your security perimeter. Your customer trust is not going to get compromised. We are enabling that with this concept called ‘Snowpark Container Services’. Snowflake has also built Document AI, a large language model (LLM) specifically designed to solve the problem of deriving structure from unstructured documents. Here, a company can take all the service reports and the Document AI will run on the service reports and give you an ability to ask questions on that data. We have also partnered with companies like Nvidia, where we are bringing their LLM to our customers. Snowflake has also acquired Neeva, the company founded by Sridhar Ramaswamy. They have built LLMs for AI-powered search. They are leveraging all of that to bring AI to every aspect of Snowflake.

What kind of traction do you see for such technologies in India?

We look at the market from three different segments. The first lens that we have is for startups and digital natives. Food delivery firm Swiggy is a great example. They are right on the top of the pyramid in terms of scale and success and what they have done as a company. When you open the Swiggy app, the restaurants, food menus, and the timing that you see about when the food is going to be delivered to you, all of it is powered through AI and due to (the Swiggy) and Snowflake partnership. We are also working with Kissht, the payments and lending fintech player. They have a brand called Ring to provide consumer lending to millions of customers and they are leveraging Snowflake’s technology. We also work with the logistics company Porter. One thing that is common across all these companies is they're using data in a very strategic manner to differentiate from their competition.

The second segment that we are looking at includes the traditional enterprises in India in areas such as automobile, pharmaceutical, retail, banking and insurance The opportunity there is massive because all of these companies are getting competition from these new startups. For example, if you are a retail company, there is an online retailer trying to eat into your market share. That is putting pressure on these traditional enterprises to innovate, transform and start leveraging data in a very strategic manner. We are really excited about the opportunities in the pharmaceutical space from expediting the drug discovery process to helping the companies, bring the services faster to the customers. The third area is the government which is very strategic; from the long-term presence that we intend to have in this market.

What is the plan to partner with the government organisations?

We completely support the government’s vision of ‘Make AI in India and Make AI work for India.’ We have already deployed our service in India for government agencies on Amazon Web Services (AWS) as well as Microsoft Azure. We see there are two (main) opportunities for us to partner with the government. The first is data collaboration. We do believe that there is immense value for different government ministries and agencies to collaborate with each other in a secure and real-time fashion. For instance, if you're in the infrastructure ministry and I am on the weather or consumer side, I can share data with you to understand consumer traffic movement. This would help you to build the infrastructure better. The governments in many developed economies are already collaborating with Snowflake, to support them in their decision-making. The other opportunity we see in the government sector is the use of generative AI and LLMs for citizen services and everything else the government is doing. We are at an early stage and are engaging with the (government) agencies. We are also adding more people in India and have expanded our presence within the country exponentially to serve the opportunity in the market.