HiBob takes control of telemetry to cut costs and improve signal

About HiBob

HiBob is a modern HR platform built for today’s global, fast-scaling companies. Their tools support the entire employee lifecycle, from onboarding to performance management, making them a core system for organizations worldwide.

But as HiBob scaled, so did its telemetry. With different active environments and rapidly growing traffic, its observability footprint expanded fast, and it was still paying overages every month. Costs were rising, and the team was spending too much time maintaining telemetry.

“We had rules in other telemetry systems, but they weren’t keeping up,” said Yishai Halpert, Director of DevOps at HiBob. “We were chasing developers, manually tuning logs, and trying to fix one-off issues.”

The team needed a better way to manage telemetry without compromising audit access or developer visibility.

Telemetry control starts with real-time insights

HiBob turned to Sawmills, a smart telemetry management platform that uses AI to analyze and optimize telemetry in-stream. Sawmills enables the DevOps team to take action in real time. They evaluated several solutions, but Sawmills was able to provide AI insights with one-click actions, along with a powerful pipeline to manage the flow of telemetry. With Sawmills, HiBob is able to:

1. Route 100% of production logs to S3, maintaining auditability without incurring ingestion or indexing costs.
2. Sample or drop high-volume, low-value logs instantly and automatically.
3. Filter out noisy attributes that were auto-instrumented but offered no real insight.
4. Aggregate repetitive logs to reduce clutter and improve developer signal.

The impact was immediate: Sawmills cut HiBob’s ingest volume by 40% and reduced indexed messages by another 30%. Even better, these changes didn’t require a single developer.

“Sawmills surfaces the issues and gives us one-click fixes. We’ve gone from chasing engineers to managing telemetry like a product.”

AI that keeps up with the noise

At the heart of this transformation is Sawmills Insights, the platform’s real-time AI engine. As telemetry flows through HiBob’s pipeline, Sawmills continuously analyzes it, flagging patterns, volume spikes, redundant structures, and then recommends specific actions like filtering, sampling, or transformation. For example, Sawmills helped identify some big problem areas:

1. One third-party integration was responsible for over 10% of total log volume. Sawmills flagged the pattern and offered a sampling rule.
2. A noisy internal service accounted for nearly half of the dev environment logs. Sawmills identified and filtered the source, cutting the volume in half.
3. Sawmills Insights detected overlaps and removed unnecessary fields, making logs more readable and cheaper to process.

“The value isn’t just cost, it’s clarity. Sawmills helps us filter out the junk and focus on what matters. We’re not just spending less, we’re debugging faster.”

^{Yishai Halpert, Director of DevOps}

Building a smarter AI SRE 

Trailblazing teams like HiBob are demonstrating how impactful AI can be in transforming incident management. They recently began an initiative to build an AI-powered SRE agent that they call Sentinel, designed to assist developers during incidents. Its goal is to enable semantic search across logs, so engineers can find relevant data even if they don’t know the exact keyword to search for. Sawmills is powering that foundation.

By using Sawmills to filter, clean, and enrich telemetry, HiBob is preparing its data for embedding, a process that transforms logs into vector representations of meaning. “Other systems can only do an exact-match search. But incidents don’t come with clear labels,” said Halpert. “With Sawmills, we’re making our SRE tooling smarter. Developers will be able to ask questions like ‘slow transaction’ and find issues tagged as ‘query lag’, even if the terms don’t match.”

This next step in HiBob’s observability strategy shows how smart telemetry management goes beyond reducing cost, to helping companies unlock new capabilities.

The results of smarter telemetry management:

• With Sawmills, HiBob has achieved:
• 1. 43% reduction in ingest volume
• 30% reduction in indexed messages
• Logs prepared for AI-powered SRE
• Cleaner telemetry and faster root case analysis
• Elimination of observability bill overages

HiBob’s journey with Sawmills reflects a broader shift in how modern DevOps teams manage telemetry. By bringing AI into the data stream, HiBob moved from reactive cost control to proactive optimization. Telemetry is no longer a source of friction; it’s a source of insight. With Sawmills, HiBob has not only cut costs and cleaned up its data but also laid the groundwork for smarter incident response and future-ready observability.