With AI revolutionising business everywhere, companies are racing to get smarter about how they manage and democratise their data.
Snowflake’s Chief Information and Data Officer, Sunny Bedi, and Sanjay Deshmukh, Senior Regional Vice President for ASEAN and India, sat down with Frontier Enterprise to discuss how the company is enabling different personas — from data scientists to business users — to interact with data using AI-powered tools.
In this interview, Bedi highlights Snowflake’s internal use of its platform, the shift to a consumption-based model, and lessons learned from his time at Nvidia as both companies navigated rapid scaling and innovation, with Deshmukh providing additional insights on data collaboration and unstructured data use cases.
Frontier Enterprise: As AI transforms business, the way different personas interact with data is becoming increasingly important. What are the key personas at Snowflake, and how is your platform addressing data democratisation?
Sunny Bedi: The key personas are data engineers, data scientists, and BI engineers. But the real value comes with business users like those in finance or sales. In the past, they would ask initial questions and then rely on the BI team for follow-ups, creating delays. It was like reading a newspaper — it tells you what happened, but you want to ask “why” multiple times to dig deeper.
We’ve solved this by leveraging AI in our platform. With tools like co-pilots and AI analysts, users can now interact with data using natural language. For example, a finance person can ask, “How many incidents were open between 5 PM and 8 AM?” and get a quick answer like “1,000.” They can follow up with, “How many for Singapore?” and get the relevant data immediately. The system can even go further by answering questions like, “Of the unresolved cases, what are the common issues?” and present all that data quickly.
It’s essentially a SQL query in natural language, converting English into SQL and providing results in real time. Users can also request trend data and anomaly detection, and the system will show them if something deviates from the norm.
Sanjay Deshmukh: We’ve extended this to unstructured data as well. This allows businesses to apply AI to documents like legal contracts, turning unstructured data into a format that’s easy to query and analyse.
Sunny Bedi: When it comes to unstructured data, we see this as a huge use case. For example, customers often have contracts stored in PDF formats that grow over time due to amendments. With AI, we can decode this unstructured data, transforming it into a structured format that allows users to ask questions and get real-time answers.
I recently visited a large insurance company in India that manages millions of documents. Their analysis processes, which used to take weeks, can now be done in minutes with AI.
Frontier Enterprise: With companies starting their AI journeys, ensuring data quality and organisation is crucial. What recommendations do you have for customers on their data strategy before embarking on AI?
Sunny Bedi: Yes, the foundation is to bring all your data into one logical place, which is Snowflake. Don’t have your data scattered everywhere — bring it into our platform, whether it’s structured, semi-structured, or unstructured. We provide governance and guardrails like role-based access control (RBAC).
Secondly, once your data is in Snowflake, there’s no need to take it out. In the past, data scientists had to move data out because they didn’t have the capabilities we now offer. Today, we provide all the tools and compute, like Snowpark and Cortex, so you can keep your data within Snowflake and still perform AI or machine learning tasks.
Some customers started their data transformation journey with legacy tools that required data to leave Snowflake. Now, with our capabilities, they’re transitioning away from those tools and using Snowflake’s features to keep their data and compute in one place.
Sanjay Deshmukh: One of our customers, Ayala, was recognised for their work in data collaboration. Another company, Temasek, was shortlisted in the same category. Both companies bring their data together and apply governance, but they also enhance their data by incorporating external sources. This approach gives them a 360-degree view of their business, allowing for better AI insights.
Sunny Bedi: Exactly. The key is to build a strong data foundation by bringing all your data together. You can also collaborate by incorporating external data, or even monetise the output of your machine learning models. We have a data marketplace where companies can share enriched data securely. For example, a consumer-packaged goods company might offer enriched data that other businesses in the marketplace can leverage.
We don’t want them to rely on traditional methods like FTP. Instead, we offer a secure way for data to move seamlessly in and out of their Snowflake instance, allowing them to apply AI and build applications on top of Snowflake.
Frontier Enterprise: How do you use Snowflake within your organisation, and when did the integration of AI begin? Was it a recent development or an ongoing project?
Sunny Bedi: Snowflake is a 12-year-old company, so we’re relatively new. Unlike many other CIOs I speak with, who are still dealing with legacy systems and transitioning to the cloud, we were born in the cloud. That eliminates a lot of the challenges, as I don’t have to manage on-prem systems, data centres, or infrastructure. Everything is in the cloud for us, so that removes a significant portion of the headaches.
As both the Chief Information Officer and Chief Data Officer, I also serve as customer zero for everything we build, develop, and innovate. We implement new technologies internally first, operationalise them, and ensure that every user within the company can provide feedback to the engineering and product teams. This helps refine the technology for enterprise use.
The result is a reference architecture that Sanjay and others can share with customers, showing how we use Snowflake. For example, we’ve provided Sanjay with apps demonstrating how our technologies come together to apply AI. When we approach customers, this reference architecture enhances their experience.
While I’m technically customer zero, all of our 7,500 employees use the product internally as part of our data-driven culture. It’s vital for every department — sales, marketing, finance — to have seamless access to the tools and data they need. My job as CIO and CDO is to make that process as smooth as possible.
Of course, we also use third-party SaaS applications for certain tasks. For example, if we need a procurement tool, we won’t build that on Snowflake — we’ll buy a solution that’s already available in the marketplace.
Frontier Enterprise: Your CRM, HR, and finance systems too, I assume?
Sunny Bedi: Exactly. But we treat data the same way we treat security. When we implement these applications and build the necessary automation, we ensure that all relevant data flows into Snowflake in real time as part of the automation process.
Frontier Enterprise: Was it a very different experience coming to Snowflake from Nvidia?
Sunny Bedi: Yes, we’re a consumption-based company, so we only recognise revenue based on consumption. It’s similar to how you pay for electricity — you only pay for what you use. There aren’t many SaaS applications that can handle revenue recognition for a consumption-based model. I’d say we’re quite unique in this regard.
This model presents a lot of challenges in terms of financial business processes — how we close the books, how we close the quarter, and how we recognise revenue. Traditional ERP systems and SaaS applications are built around subscription models. They sell you a fixed number of licences, and then come back in two years for a renewal. There’s no focus on optimising that for the customer.
In contrast, we provide a consumption-based model where customers can optimise their usage, just like managing electricity or air conditioning at home. You don’t leave your doors open and crank up the AC — you optimise. We provide the tools to help customers optimise, but all of that has implications for how our financial systems operate in the back end.
Frontier Enterprise: Was the transition from Nvidia to Snowflake a big change for you?
Sunny Bedi: I was at Nvidia for 12 years, and I had two different roles there — first in operations and product introductions, and later leading the systems team. When I joined Nvidia, it was a smaller company, and I had the opportunity to help it scale significantly.
At Snowflake, I joined when the company was still relatively small, but it has grown rapidly since then. The scaling challenges between both companies are quite similar, and I’ve brought the lessons I learned at Nvidia to Snowflake.
It’s all about building scalable processes, automation, and infrastructure, so that teams — whether in sales or product development — can grow without being held back by systems or processes. That mindset has been crucial at both companies, ensuring that teams like Sanjay’s can operate smoothly with the right data and systems in place.
Nvidia’s success didn’t happen overnight; it was the result of long-term investment. We’re hoping to replicate that trajectory here at Snowflake.
