Snowflake Startup Spotlight: Contextual AI

Welcome to Snowflake’s Startup Spotlight, where we learn about awesome companies building businesses on Snowflake. In this edition, we talk to Douwe Kiela, the CEO and co-founder of Contextual AI, a startup that helps companies build highly specialized, production-grade AI agents. These agents are capable of reasoning over enterprise data using retrieval-augmented generation (RAG), making them uniquely able to very accurately understand the context of a business. 

What inspires you as a founder?

The potential of AI truly inspires me. I believe it can revolutionize the world and solve critical global problems.

What problem does Contextual AI aim to solve?

While it’s very easy to put together a compelling AI demo, it’s very difficult to get AI systems into production. This is especially true for enterprises, which demand high levels of accuracy, auditability and security. 

Having led the research team at FAIR that wrote the 2020 paper on RAG, I knew that RAG would be an important component in building AI for enterprises. However, as RAG gained widespread popularity, I could see that the way people were implementing the concept was suboptimal. I started Contextual AI to provide a better way for these enterprises to leverage AI, enabling them to achieve the outcomes they need more quickly.

What’s the coolest thing you’re doing with data? 

Currently, people implement RAG pipelines that look like Frankenstein’s monster. They have to stitch together extraction, embedding, retrieval and language models that aren’t designed to work well together. This is why most companies fail to reach production quality with their RAG solutions 

Our platform is different: it tightly integrates and jointly optimizes all the components needed for RAG workflows. This approach enables our customers to achieve much higher accuracy, helping them get these solutions into production faster than their competitors. 

How do security concerns impact the deployment of your platform?

Enterprises are very security conscious when it comes to their data and their AI projects. The Contextual AI platform can be deployed into a customer’s virtual private cloud, but this requires their engineering teams to set up and manage the infrastructure that the platform runs on.

When they deploy Contextual AI as a Snowflake Native App, they get the peace of mind that comes with running the platform and the data processing inside their own Snowflake environment, while Snowflake manages the infrastructure complexity, including server management, scaling and maintenance.

How has the Snowflake Native App Framework affected your startup's growth?

Large enterprises need to trust and verify that vendors will properly handle and protect their data. Even with our own certifications, the security review process during deal cycles can be lengthy and really slow down momentum. 

Offering our solution as a Snowflake Native App accelerates the sales cycles because companies can deploy the fully configured app into their existing Snowflake environment, which has already gone through extensive security reviews.

How does being on Snowflake Marketplace affect customer acquisition for you?

We’re very excited to take part in the Marketplace Capacity Drawdown (MCD) program. Allowing enterprises to pay for our platform on demand with their existing Snowflake credits will drastically reduce our sales cycles. 

What’s your vision for where AI is headed?

AI agents have been a huge focus this year, but most of the products being released serve general-purpose use cases like web search. We believe what enterprises actually need are agents that are optimized for industry-specific workflows and understand the context of internal data and business processes so they can be much more accurate. 

Our company is focused on addressing that need, providing a platform for building specialized RAG agents that serve use cases such as investment analysis, engineering design support and technical research — areas where general-purpose agents have largely fallen short.

What advice would you give to others considering building their applications on Snowflake?

I would say, do it! Building on Snowflake has huge benefits for getting your product to market. Beyond what I’ve already mentioned — like simpler security reviews, faster sales cycles and the MCD program — it’s amazing to have Snowflake’s global sales team working with you. 

Learn more about Contextual AI, and how it helps organizations build and deploy production-grade RAG agents that are highly auditable and specialized, at contextual.ai. You can also try the Contextual AI Platform for yourself on Snowflake Marketplace.