PagerDuty Launches Paige AI Agent and Renames Its Core Product PD Reliability Platform

The company also opened PD Labs, a programme that gives design partners early access to features still in development and direct contact with the engineers building them. Paige and the PD Reliability Platform are generally available now.

Why PagerDuty Says It Needs an Agent

PagerDuty presents the launch as a response to a growing reliability problem. AI coding tools are writing more of the software that goes into production, often with limited human review, and applications built on AI models fail in new ways. Chief Executive John DiLullo gave model drift, token runaway and leaked personal data as examples. He said the only way to stay resilient is "to fight fire with fire, using AI itself as part of how we keep systems running."

Customers decide team by team how much Paige can do. Where a team wants a person involved, the agent speeds up triage but holds any fix until someone approves it. It acts on its own only in well-understood, low-risk situations the team has explicitly authorised. PagerDuty said in March that its SRE Agent would begin joining escalation policies as a "virtual responder". Paige, described as the next generation of that agent, can now be placed on call in the same way, so some lower-priority incidents may never reach a human.

Paige attaches supporting evidence and a confidence level to each recommendation. Resolved incidents go into a shared memory that the agent checks when similar failures occur, and PagerDuty argues this is where most of the long-term value lies. In a benchmark report released with the launch, the company said Paige beat leading AI coding agents at diagnosing simulated IT incidents. That claim comes from PagerDuty and has not been independently verified.

Console, Chat Integration and Labs

Incident commanders and team managers get a new PD Reliability Platform Console. It shows incidents, on-call rotations and AI agents in one place, including which agents are active and what each is doing, and Paige's suggested fixes can be applied from the same screen. Responders can also run an incident from start to finish in Slack or Microsoft Teams. A trigger opens a dedicated channel and a conference bridge automatically, and the post-incident review is generated as soon as the incident is resolved, with follow-up tasks already assigned owners and due dates.

Canadian telecoms group Telus has connected the platform to its own internal incident tool. Kulvir Gahunia, director of Telus's Site Reliability Office, said PagerDuty's Scribe Agent now runs through every incident so responders "don't have to stop and document what is happening."

PD Labs starts with three preview experiments. Topology builds a live map of an organisation's services from its code, infrastructure and runbooks, without needing a configuration management database, and works out how far a problem could spread. Deep Investigation lets Paige test several root-cause hypotheses at once and show its reasoning as evidence comes in. SRE Workbench gives people and agents a shared workspace, and anything they find there is saved to Paige's memory. SRE Workbench is available to try now.

Pricing is based on operational scale and includes a starting allowance of ingested events, automation actions and user seats. Customers who go beyond it can buy flexible credits without forecasting usage in advance.

What It Means for IT Leaders

The launch follows a wider shift among enterprise software vendors from AI assistants to agents that take action within limits set by customers. Snow Tempest, research manager for IT service management at IDC, said IT teams want measurable returns from their spending on AI agents, such as better uptime and time returned to responders who would otherwise be "chasing logs and assembling the right people for an incident." For companies running AI-heavy software, the key decision is how much autonomy to give such agents, and PagerDuty's team-by-team permissions are designed around that choice.