Amazon's AI Cuts Database Fix Time by 78% for 140 Million Users

Cornerstone OnDemand, which provides workforce readiness solutions to 140 million users worldwide, had a problem: when their databases had issues, it took engineers up to 45 minutes to figure out what went wrong. They manually checked logs, ran queries, and coordinated across teams. That's like having a car that breaks down and waiting 45 minutes just to diagnose the problem before anyone can fix it. So they built Orion AI, a system that uses Amazon Bedrock and Strands Agents to create a team of specialized AI helpers that work together to diagnose and fix database issues. The result? Diagnosis time dropped to just 10 minutes—a 78% reduction. A three-person team built it in six months.

Orion AI works like a smart assistant that coordinates several experts. One AI agent monitors infrastructure, another diagnoses database problems, another handles lifecycle tasks, and so on. When an issue arises, the main AI (the meta-orchestrator) delegates tasks to the right agents. Engineers interact through a web app, asking questions and approving actions. The AI not only finds the root cause but also recommends a fix and creates a Jira ticket for the on-call engineer. This turns a four-step manual handoff into a single interaction. It also cuts alert fatigue: redundant alerts are reduced by 65%, so for every 10 alerts, only 3-4 reach engineers. That means less noise and more focus on real problems.

Beyond diagnosis, Orion AI automates database lifecycle management. Tasks that used to require over 10 manual steps—like connecting to databases, running queries, and updating status—now happen with a simple natural language prompt. The AI identifies the right tools, runs queries, validates results, and returns a unified response. Engineers just prompt and validate. This frees them from repetitive work and lets them focus on more strategic tasks. The system also provides real-time monitoring, removing a 15-minute reporting lag and giving continuous visibility across systems.

The key takeaway? AI can be a powerful teammate for IT operations, speeding up fixes and reducing burnout. While this is a specific case, it shows how AI can make software more reliable and support teams more efficient. For everyday users, that could mean fewer disruptions and faster resolutions when things go wrong.