Snowflake acquires TruEra to deliver LLM observability inside data cloud

Join us in returning to NYC on June 5th to collaborate with executive leaders in exploring comprehensive methods for auditing AI models regarding bias, performance, and ethical compliance across diverse organizations. Find out how you can attend here.

Today, data ecosystem giant Snowflake announced it has entered a definitive agreement to acquire TruEra, an AI startup providing tools to test, debug and monitor machine language (ML) models and large language model (LLM) apps in production.

While the terms of the deal remain under wraps, the acquisition is expected to bolster Snowflake’s effort to give its customers an end-to-end platform to build generative and predictive AI applications using the data hosted on their Snowflake data cloud. The company already provided models and tools to build gen AI apps and is now moving to add the capabilities that will ensure the apps built are effective and trustworthy.

This is Snowflake’s sixth major investment to bolster the capabilities offered on its data cloud and the third major move in the observability space. Before this, it invested in monitoring solutions Observe and Metaplane.

What does TruEra bring to the table?

Founded in 2019 by William Uppington, Shayak Sen and Anupam Datta, TruEra tackles the “black box” problem in machine learning by providing a comprehensive observability platform that evaluates, debugs and monitors machine learning models and apps across the full lifecycle from development to production.

The company started with traditional machine learning observability but evolved over the last few years to cover large language models, enabling users to assess the input, output and intermediary results of LLMs — essentially checking whether they produce accurate and relevant results and guard against risks such as hallucinations, unfair bias, toxicity and more. It also started providing detailed insights with specific root cause analysis to help teams improve the performance of their models.

Now, by joining Snowflake with this acquisition, TruEra’s technology will power the data cloud, giving users building next-gen AI apps, including question answering, summarization, and RAG chabots, a tested solution to check for performance gaps. The startup’s engineering and executive team, including all three co-founders, will also join the Sridhar Ramaswamy-led data giant.

“TruEra’s capabilities complement the AI and ML data governance functionalities we already provide in the AI Data Cloud. Snowflake provides deeply integrated capabilities to ensure the accuracy and trustworthiness of the data used to supplement and train models, and the observability technologies developed by TruEra will complement and round out that story for AI,” Snowflake wrote in a blog post.

In a separate post, the founders of TruEra added that the core elements behind the company – world-leading AI research, hands-on experience with building and deploying ML, domain expertise with specific verticals and a shared passion to make AI adoption effective and safer – will remain key in the next phase of the journey with Snowflake.

Race to build end-to-end capabilities for AI

While it remains unclear how exactly Snowflake will integrate TruEra’s AI observability capabilities into its platform, the deal is expected to further support the company’s mission to deliver end-to-end capabilities for building high-performance AI apps with data stored in the data cloud.

Before focusing on the reliability side of AI with TruEra, Snowflake took several steps to simplify LLM app development on its data cloud. The company launched Snowflake Cortex, a fully managed service to build LLM apps, last year and has since enhanced it with several open models, including those from Reka and Mistral. A few weeks back, it also expanded Cortex with support for Arctic, an open LLM designed for complex enterprise workloads such as SQL generation, code generation and instruction following.

Snowflake’s focus on AI has grown since Ramaswamy took over as the CEO. Notably, the AI-first approach also enables the company to better take on rival Databricks, which started with a focus on AI apps and continues to build up its data intelligence platform with a similar mindset of building end-to-end capabilities. The company acquired MosaicML last year and has since developed its AI offerings around it.


Share This Article
Leave a comment