Monte Carlo raises $25 million for AI that monitors data reliability

San Francisco-based data reliability startup Monte Carlo today announced that it raised $25 million, bringing the company’s total capital raised to date to over $40 million. Monte Carlo says the proceeds will allow it to foster its community of users and further develop its data and analytics products as it looks to expand the size of its workforce.

It’s estimated that the average company spends upwards of $15 million annually tackling periods of time where data is missing, broken, or otherwise inaccurate. Around 93% of organizations who lose servers for 10 days or more during a disaster file for bankruptcy within the next 12 months, while 43% never reopen and 51% close within two years.

Barr Moses, former VP of customer operations at Gainsight, cofounded Monte Carlo in 2019 with Lior Gavish, an ex-SVP of engineering at Barracuda Networks. Both were struck by what they perceived as an ease of use problem when it came to tools for identifying and resolving infrastructure issues: While these tools were widely available, they didn’t offer a simple way to guarantee the validity of data flowing through pipelines.

Monte Carlo uses AI to infer and learn what a company’s data looks like, proactively identify downtime, assess its impact, and notify employees who might need to know. The platform can automatically spot the root cause of downtime and show data dependencies in one place. Moreover, it offers a code-free implementation for out-of-the-box coverage with existing data stacks, with a single view of data health spanning data lakes, warehouses, business intelligence tools, and catalogs.

“[We] automate a traditionally manual process of data validation and monitoring that relies on time-intensive threshold setting,” a spokesperson explained. “[Monte Carlo] uses machine learning to take a historical snapshot of data assets to prevent ‘bad data’ from corrupting otherwise good pipelines. Through this approach, we take stock of data assets and use machine learning to determine which ones are most ‘critical’ — i.e., which ones are most widely used, how many people are using them, and how they are using them. We also use various AI anomaly detection techniques to benchmark historical data, metadata and patterns in the customer’s environment, and then identify substantial deviations from these benchmarks that could indicate the presence of ‘bad data.’”

Importantly, Monte Carlo monitors data at rest and doesn’t extract it from datastores. Thanks in part to this, it’s one of the few data observability platforms to achieve SOC-2 compliance, according to CEO Moses. (Developed by the American Institute of Certified Public Accountants, SOC-2 is designed to assess the security of service providers storing customer data in the cloud.) The training dataset for Monte Carlo’s algorithms only includes information from individual companies so as to prevent leakage across customers.

“In recent years, solutions like Splunk and New Relic emerged to prevent application downtime. In 2021, the same thing is happening with data. As companies ingest more and more data to fuel decision making, ‘data downtime’ has emerged as the biggest threat to businesses everywhere,” Moses told VentureBeat via email. “Data observability is the first end-to-end, ML-driven approach to understanding the health of data at each stage of its lifecycle. Monte Carlo is excited to continue pioneering this new category, and in the process, empower companies to finally achieve data trust.”

Twenty-five-employee Monte Carlo — which has offices in Canada, South America, and Israel and counts among its customers teams at Eventbrite, Snowflake, and New Relic — says that revenue doubled every quarter of 2020. The company now manages the reliability of over 20 petabytes of data.

Repoint Ventures and GGV Capital led Monte Carlo’s latest funding round, which had participation from Accel. It comes after the company’s $16 million series A round in September.