The future of observability won’t be one proprietary AI agent. It will be thousands built by teams.

The future of observability is shifting towards a decentralized model where thousands of AI agents, tailored to specific organizations, teams, and problem domains, replace the notion of a single proprietary AI agent. These agents are becoming the primary interface for observability, capable of querying telemetry, identifying patterns, and generating hypotheses, thus handling much of the investigative workload traditionally managed by human engineers. This transition emphasizes the need for openness, allowing teams to build agents around existing systems and processes rather than conforming to a vendor's framework. As AI agents become more integral, they increase the demands on underlying systems due to their ability to pursue multiple hypotheses simultaneously, necessitating robust data infrastructure. Despite these advancements, human engineers remain central to the decision-making process, with agents providing context and evidence rather than making autonomous decisions. This collaborative model fosters a shared workspace where investigation artifacts can be preserved, reused, and refined, ultimately creating a comprehensive body of operational knowledge for future reference.