A series of technology outages has disrupted banking, airlines and enterprise software in recent days. The Federal Aviation Administration grounded flights nationwide on Nov. 13 after a router replacement, and JPMorganChase said a software bug delayed transactions on Nov. 16, PYMNTS reported this week. Microsoft said a bug introduced in July left more than 80% of its Microsoft 365 gateway servers unresponsive during a global outage that began Nov. 12. The incidents were unrelated.
Companies are using artificial intelligence to limit that kind of damage. These AI systems monitor infrastructure, flag unusual behavior and help engineers resolve incidents before they spread. Among U.S. companies with at least $1 billion in revenue, 61% of the heaviest AI users apply AI to incident detection and response, compared with 32% of the lightest users, PYMNTS Intelligence found. T-Mobile, MSCI and Microsoft have each published results that show what that looks like in practice.
T-Mobile’s AutoPilot Halves Network Fix Times in Testing
T-Mobile, the wireless carrier, said in September that its AutoPilot system made real-time network adjustments in about half the time in testing. AutoPilot adds AI to the carrier’s self-organizing network. When a cell site goes offline, AutoPilot determines how neighboring sites should adjust their coverage, which closes service gaps while engineers fix the problem. T-Mobile is expanding AutoPilot nationwide.
T-Mobile’s self-organizing network has already done that work in the field. During Winter Storm Fern in January, automated changes kept cell sites online for more than 250,000 additional minutes across more than 30 states, and the carrier made more than 30,000 antenna adjustments. Combined with generator deployments, that work restored coverage to 68% of affected customers within one hour and 98% within eight hours, T-Mobile said.
MSCI Team Cuts Reported Incidents 77%
MSCI, the financial data provider whose indexes are benchmarks for $21 trillion in assets, uses AI agents from the startup Resolve AI to take the first pass at investigating incidents. Reported incidents fell 77% on one MSCI infrastructure team and 76% on another, Resolve AI said in an October case study. Client cases fell 46% from their peak by June.
MSCI deployed Resolve AI across more than 12 monitoring tools. The AI examines alerts, connects related failures and determines which infrastructure component needs attention. In one incident involving failed index calculations, the AI connected five technical signals and traced the problem to an outdated server address in 14 minutes. Similar investigations had taken MSCI engineers three to four hours.
More than 80% of MSCI’s critical incidents now get AI triage before engineers take over.
Microsoft Mitigates High-Severity Azure Incidents 50% Faster
Microsoft’s Mission Critical Services for Azure Platform team has cut the time needed to mitigate high-severity incidents by more than 50%, Microsoft said in a September case study. The team handles incidents for Microsoft’s largest Azure customers. Its AI agents assemble incident histories, identify likely causes and prepare root-cause reports that used to take engineers up to two hours. Low-risk incident volume has fallen by as much as 40%. The figures come from Microsoft’s internal data for the 12 months through June.
In one case, an Azure Key Vault outage grounded flights at a large U.S. airline, and establishing the business impact took about five hours. When a similar issue came up, Microsoft’s agents identified the dependency within minutes, the case study said.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.