Harness, a software delivery startup that provides AI tools for developers to update and monitor applications, today announced the launch of a "humanized change agent" designed to help teams resolve incidents faster without needing to add extra personnel.
The company states that AI for site reliability engineering (SRE) integrates key components of response engineering by capturing context, coordinating actions, and providing investigative assistance. The new product, AI Scribe, is built precisely for this purpose.
When AI is applied to SRE, analysts and engineers are often likened to detectives. They examine incident reports, logs, and diagnostics, much like investigators scrutinizing a crime scene. It's a chaotic mass of information, typically a complex cascade of patterns that is difficult to unravel without a foundational understanding of the system and the basic knowledge of what, where, why, and who is responding, and when.
Furthermore, sometimes the most critical "clues" originate from human complaints or early reactions. For instance, someone might report, "A customer said the checkout button froze after updating their cart," or "Service X felt slow an hour before this started," or "Didn't we flip a flag for the recommender earlier today?" or "This is only happening in the US-East cluster."
Information emerging from Slack, email, or other messaging systems can quickly shape the investigation before anyone even begins reviewing reports, logs, or launching dashboards.
Harness explains that AI Scribe operates as a conversational interface. It captures operational signals from the team's natural discussions around affected services, dependencies, symptoms reported by customers, proposed or conflicting theories, and key sequential clues ("just before that..."), and attempts to synthesize them.
It investigates these "clues" as human signals and provides a change map, utilizing feature flags, profiles, infrastructure updates, and other elements naturally associated with the operational machine structure. The goal is to present a hypothesis that the team can comprehend.
Rather than presenting wild guesses to the team, it offers clear, explainable insights accompanied by reasoning and supporting data. This enables teams to quickly validate or discard ideas and move forward.
Harness expresses its ambition for AI Scribe to become an incident responder that delivers functional intelligence, providing a faster path to the truth, not merely summarizing events.
