Observe unveils AI-powered agents to speed troubleshooting

Cloud observability provider Observe today introduced updates to its product set with Project Voyager, a SaaS offering that now includes an artificial intelligence (AI) investigating tool and capabilities to monitor application performance as well as to query Snowflake data storage services.

Observe, founded in 2017, set out to consolidate the capabilities of multiple enterprise IT tools into one platform. With the release of Project Voyager, Observe CEO Jeremy Burton says the company is completing its initial vision to build out a unified observability platform that integrates log analytics, tracing, and monitoring. As applications have grown more sophisticated, it has become more challenging to aggregate, normalize, and analyze the data, especially when incidents arise.

“We now do log analytics, tracing, and monitoring all in one integrated product. The APM pillar of the Observe puzzle is the final piece,” Burton explains. OpenTelemetry provides a common format for collecting and transmitting telemetry data, which allows organizations to use multiple tools to collect data without having to reinstrument their applications. And it is becoming the industry standard for collecting telemetry data from applications in an open, vendor-neutral way.

“This allows our platform to provide a comprehensive view of application performance, in addition to the existing log analytics and monitoring features. Observe now offers customers a single pane of glass for troubleshooting complex, distributed applications—which is a critical need in modern, cloud-native environments,” Burton says.

Also new with this release is Observe’s AI Investigator, which orchestrates a network of domain-specific AI agents to assist engineers in quickly identifying and resolving issues. The AI Investigator works with AI agents that are configured to learn specific tasks or domain knowledge, such as accessing runbooks or deeply understanding Kubernetes, AWS, or Github. An AI planner drives the troubleshooting workflow for an on-call engineer, gathering necessary data to triage a problem before having to contact the Kubernetes expert, for instance, to simply gather the relevant Kubernetes data.

“The big thing for this release is a move toward trying to impact the troubleshooting of the application itself. If you think about it, the average time to resolve a major incident in the industry is about two and a half hours. The first hour is just gathering information. We thought this would be a good use for AI if you are the on-call engineer. Instead of getting people out of bed at 2 am, you could call on virtual versions of those people to gather all the necessary troubleshooting data,” Burton explains.

Industry watchers agree that it makes sense to apply AI to observability platforms.

“There is an immense opportunity in leveraging AI for the modern observability industry,” said Kate Holterhoff, senior analyst at RedMonk, in a statement. “Observe’s AI-powered investigation features are a promising addition to this growing market.”

Also with this release, Observe added Snowflake Observability, which is now available in the Snowflake Marketplace. This integration allows developers to gain insights into query performance and application health without moving data outside of Snowflake, which will ensure security and efficiency, according to Observe.

Observe is a SaaS platform, and customers deploy Observe agents to collect telemetry data. The agents can collect data from a variety of sources, including infrastructure such as Kubernetes, databases such as MongoDB or Snowflake, and other applications. The agents collect time-series data, logs, traces/spans, and performance data from these various sources and send the data to Observe’s platform. Observe then takes the raw telemetry data, curates and normalizes it, and structures it to make it more easily navigable and usable for troubleshooting by customer teams.

Customers access the platform with a web-based user interface, which provides observability and troubleshooting tools and capabilities. Primarily Observe is used by DevOps teams, site reliability engineers, and engineering teams, but it can be easily accessed by a variety of roles within an organization.

With the Project Voyager news, the company also announced an additional $20 million in Series B funding, bringing its total to $145 million.