Snowflake is a data platform that harnesses the power of the cloud and allows organisations to explore, share and unlock the true value of their data.

Retailbiz recently spoke to Snowflake global head of retail, Rosemary DeAragon, who advises retailers on how to best leverage the Snowflake platform for Artificial Intelligence (AI) and more generally, data and analytics.

“There’s been a range of trends in the retail space but one of the most hyped trends is around generative AI and large language models (LLM), as well as retail media and data monetisation to create new revenue streams,” DeAragon told Retailbiz in a recent interview.

“The Snowflake platform has features that connect companies and allow them to collaborate. For example, our data sharing feature allows consumer products companies and their retailer partners to query each other’s data by simply setting permissions on that data.”

In the past, organisations would need to build an API or use outdated methods of moving data such as emailing spreadsheets. Since retailers transitioned to using the cloud, Snowflake was able to facilitate data sharing, regardless of the tech stack, as a multi-cloud business across Google, Microsoft and Amazon.

“For retailers, the Snowflake platform is so powerful because it allows them to source datasets from any other company, no matter which cloud they’re on, and monetise their own data,” DeAragon said.

Leveraging generative AI beyond basic functions

With increasing attention on the latest developments in AI, particularly generative AI, DeAragon went on to explain some use cases for retailers building the technology into their platform.

“The first, and probably the most obvious, is using gen AI as a shopper assistant. The great thing about LLM is that it generates content in human readable terms, and you can bring that open source LLM, or use one from open AI, to bring into your internal platforms and allow users to ask questions about your platform,” she said.

“For example, US grocery delivery and pick-up service, Instacart has launched ‘Ask Instacart’ (abbreviated to AI). With this tool, you can integrate LLM into your app for customers to start asking questions, such as ‘What snacks will my kids love?’ or ‘What can I cook for someone who is gluten-free?’ and gen AI will help build a list of suitable items.”

Acknowledging this as the most basic way to use gen AI or GPT, the newer version provides the ability to build plugins.

“For example, if you want to make a Mexican dish, it will give you ingredients and step-by-step guidance on how to make it. With the GPT plugin, you can now add those items to a cart linked to the retailer’s app. This is game-changing for retail because now you can augment the ways that customers shop and check out. In this instance, Instacart represents the retailer,” she said.

“One of the most important aspects of building something like this is ensuring that the LLM doesn’t get trained on the retailer’s internal proprietary data because once you train an LLM, you can’t untrain it.  You want to make sure the data is super secure and stays in the same place. You need to bring the workload to where the data resides. This is where Snowflake can help because we ensure the data is protected. You want the AI to come to the data, as opposed to moving your data out, and having data leave that ecosystem.”

AI has been around for some time for a range of enterprise use cases, but the generative aspect is new because it allows organisations to generate content that never previously existed from an idea to an essay.

“For retailers who have only used AI for traditional use cases, such as fraud detection, my recommendation would be to make sure that the use case has a high level of inaccuracies because LLM collects and scrapes a broad amount of content to create summaries.

“For example, Amazon is using gen AI to summarise customer reviews. When a customer is purchasing a product like a toy, they will often have a look at ratings and reviews but only pick out one or two reviews. Now Amazon is scraping hundreds or even thousands of reviews and providing the customer with a summary of the positive and negative feedback to make shopping easier for the customer.”

A cookie-less future

As cookies go away, retailers will need to rely more on first-party data and customer information on hand, which will become the ‘new gold standard’, according to DeAragon.

“Snowflake’s data clean rooms feature has the ability for customers to collaborate on first-party data without moving that data. Simply use permissions to allow people to access the data without needing to move it around or build pipelines. This feature enables retailers to collaborate and enhance first-party data and act on that data directly without having to use traditional methods,” she said.

“On the retail media side, we are helping retailers figure out their retail media strategy, bid on ad inventory, enrich their customer data and then activation and measurement, such as return on investment (ROI).”

What’s next for Snowflake?

Snowflake is making significant investments in gen AI and retailers should look forward to upcoming announcements.

“Earlier this year, Snowflake acquired search company, Neeva, powered by LLM to build conversational experiences, and last year, we acquired Applica, an AI-based text automation platform. Applica can even generate insights from unstructured text, such as photos and videos and there’s already been more than 1,000 applications built by Snowflake users,” DeAragon said.

“More recently, we partnered with Nvidia, allowing businesses to securely build custom LLM using their own proprietary data in the Snowflake Data Cloud. There’s a lot happening in the world of gen AI and I only see these use cases deepening over the next six to 12 months.”