Organizations today are dealing with more telemetry than ever before, but the quality of that data has not kept pace. One of the most persistent issues is the heavy use of unstructured logs, which are logs shipped as a single string without clear attributes, context, or fields. These logs are hard to query and difficult to visualize, slowing down engineering teams when reliability matters most.
Sawmills was built to solve exactly these kinds of problems. As the first AI-powered telemetry management platform, Sawmills analyzes your data before it reaches observability tools. The platform identifies waste, improves data quality, safeguards reliability, and enables DevOps teams to control their telemetry pipelines with far less effort. Today, we’re expanding that mission with a major upgrade to our AI Telemetry Insights and actions, which now allows users to identify and transform unstructured logs into structured logs with a single click.
Why Unstructured Logs Create Engineering Burdens
Unstructured logs are extremely common in modern applications, especially in large microservice environments where different teams and different frameworks and OSS services use different logging patterns. Unstructured logs arrive as a long, unformatted message, bundling everything together: timestamp, IP address, severity, and message content.
Because nothing is broken into attributes, these logs are harder to search, filter, and use for dashboards or alerting. This is one of the biggest sources of what we consider “low-quality” telemetry: data that exists but isn’t accessible enough to drive action.
How Teams Have Been Handling This Up Until Now
Most organizations want structured logs, but the path to getting there is manual and cumbersome. The typical workflow has DevOps writing and maintaining complex parsing rules that require writing brittle regex to force structure into string-based logs. Every variation in format introduces more work, and when code evolves, those rules must be revisited. It becomes a constant cycle of tickets, regressions, and rework. Teams run out of time, and unstructured logs remain unstructured.
Introducing Unstructured-to-Structured Automation in Sawmills
The latest upgrade to Sawmills AI Telemetry Insights eliminates manual parsing entirely. Sawmills now detects unstructured log patterns in real time. Using LLM-driven semantic analysis, the platform identifies key data elements such as timestamps, IP addresses, severity levels, request identifiers, and more. From that analysis, it generates a recommended schema and automatically creates an insight that recommends an action to automatically transform unstructured logs to structured.
There is no regex to write or maintain. No redeploy required. No ticket ping-pong. Logs that were once opaque strings arrive downstream as clean, queryable fields from the moment they are generated.
“Teams shouldn’t have to choose between moving fast and keeping telemetry usable,” said Ronit Belson, CEO of Sawmills. “This capability gives DevOps higher data quality with no manual effort. It’s a major step toward telemetry pipelines that optimize themselves.”
Raising the Quality of Telemetry Across the Stack
Unstructured logs don’t just slow down investigations; they also make them more difficult. They make every downstream action less reliable. When logs are structured, dashboards align, queries return faster, alerts become more accurate, and correlation across services becomes straightforward. Data that was once hidden in generic strings is now available to power real insights.
By automating structure, Sawmills helps teams unlock the value already inside their telemetry while preventing them from paying for data they can’t meaningfully use.
This release reinforces our goal of building intelligent telemetry pipelines that automatically detect quality issues, propose remediations, and enable DevOps to apply fixes with a click. Unstructured log detection and transformation is one of the highest-impact examples of this vision in action, and today’s update makes it available to all Sawmills customers.
Get started with Sawmills.
Unleash AI in your telemetry pipeline. If you’d like a guided walkthrough of the feature or help enabling it in your environment, we’d love to show you what’s possible. Request a demo.
