PagerDuty has launched Paige, an AI agent for site reliability teams, alongside a renamed reliability platform and an early-access programme for new tools. The practical pitch is to speed up incident diagnosis while letting teams decide how much authority the agent gets.
Watch Desk analysis
What happened
Paige can join on-call escalation policies as a “virtual responder”, investigate incidents and present recommendations with evidence and a confidence level. PagerDuty says teams can set permissions individually: people can approve proposed fixes, while autonomous action is limited to low-risk scenarios they have explicitly authorised.
The launch also brings a console for incidents, on-call rotations and agents; incident response through Slack or Microsoft Teams; and automatically generated post-incident reviews with follow-up actions assigned. SiliconANGLE reports that the platform is generally available, with pricing tied to operational scale and a starting allowance for ingested events, automation actions and user seats.
PagerDuty has opened PD Labs for early features developed with design partners. Its first previews include Topology, which maps services and helps scope an incident’s blast radius; Deep Investigation, which tests several possible causes in parallel; and SRE Workbench, a shared workspace for people and agents.
The company also says Paige outperformed leading AI coding agents on simulated IT incident diagnoses in a benchmark released with the launch. That is the company’s benchmark claim, not an independent evaluation.
Why it matters
AI can help create and change production software, but it can also add new failure modes and more systems to watch. PagerDuty is applying AI to the response side of that equation: gathering evidence, narrowing possible causes and, where teams permit it, taking action. For responders, the potential payoff is less time chasing logs and coordinating people during an outage.
The permission settings matter as much as the agent’s diagnostic ability. A recommendation that a person can inspect is a different proposition from a repair made without approval; PagerDuty is offering both modes, with autonomy bounded by what each team authorises.
Our read
This is more than an AI label on an incident dashboard: Paige can be placed on call, and the platform connects diagnosis to response and follow-up. The useful test will be whether it reaches sound conclusions quickly in real incidents, and whether its controls make the difference between helpful automation and a very fast way to make a bad day more interesting.
What to watch
- Whether customers publish results on diagnosis time, uptime or time saved for responders.
- How often teams allow Paige to act without human approval, and what they restrict.
- Whether the PD Labs previews become generally available, and on what terms.
Discussion spark: For incident response, should an AI agent be allowed to fix low-risk problems on its own, or should every change wait for a human sign-off?
Sources and evidence
- PagerDuty debuts Paige AI agent with renamed PD Reliability Platform and PD Labs – SiliconANGLE (3 November 2026, 13:00 UTC)
- Startup Athena raises $2.5M in pre-seed funding to revamp the revenue model for legal AI – SiliconANGLE (3 November 2026, 13:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.