September 2026 Summaries

PagerDuty’s 2026 report argues that AI adoption in digital operations is associated with stronger resilience and revenue growth, while acknowledging organizations’ concerns about granting autonomous systems control over production infrastructure. It presents a five-stage incident lifecycle in which AI supports detection, triage, diagnostics, remediation, and learning by reducing alert noise, routing incidents, compiling context, suggesting root causes, automating approved fixes, and producing post-incident analyses. Human teams retain responsibility for strategic judgment, novel incidents, customer communications, and approval of higher-risk actions, with AI autonomy scaled according to how well an incident type is understood. PagerDuty positions its platform and SRE agent, Paige, as tools for integrating these capabilities into a continuous learning system that retains operational knowledge from each incident. The approach emphasizes configurable governance and gradual expansion of AI authority, allowing teams to automate repetitive work while concentrating on complex, context-dependent decisions.

PagerDuty has rebuilt its Slack integration into a configurable incident-management workspace designed to reduce tool switching during outages. Its SRE Agent, Paige, can use connected systems, workflows, and historical context to help triage incidents, run diagnostic and remediation actions, escalate issues, and generate stakeholder-ready status updates, while the Scribe Agent transcribes incident calls and produces summaries for post-incident reviews. New Custom Cards allow organizations to tailor incident information by Slack channel type, giving responders, notification recipients, and stakeholders different views with relevant default and custom fields. Together, these capabilities support an incident lifecycle managed largely within Slack, from declaration and automated mobilization through diagnosis, communication, resolution, and review.

PagerDuty has introduced two intelligent Model Context Protocol tools in Early Access to help AI agents support incident response without requiring users to switch between multiple applications. Through its Remote MCP Server, supported clients can already manage the incident lifecycle, including creating, acknowledging, reassigning, annotating, and resolving incidents, while the new blast radius analysis tool uses historical incident relationships to identify affected dependencies, downstream risks, correlation strength, and potential user or account impact. A second tool generates AI-assisted post-incident review drafts using incident timelines, services, participants, and other recorded context, allowing teams to document lessons while details remain fresh. PagerDuty positions MCP as a deliberately designed interface for agents rather than a simple collection of APIs, aiming to provide operational intelligence across tools such as Slackbot while retaining human judgment and accountability for decisions.

AI-driven automation is reducing entry-level engineering hiring while increasing demand for senior engineers able to oversee complex AI systems, raising concerns about future talent pipelines, operational knowledge loss, security, and resilience. Citing survey data and industry examples, the discussion argues that organizations should use AI to remove repetitive work rather than redirect engineers into supporting AI models, allowing people to focus on system design, incident analysis, auditing, validation, and other higher-level responsibilities. It emphasizes psychological safety, blameless incident practices, mentoring, and knowledge sharing as ways to preserve and transfer expertise across engineering teams. It also advocates continuously capturing operational lessons from incidents and applying them through AI-assisted routing, triage, documentation, and automation, with PagerDuty’s Scribe and SRE Agents presented as tools designed to support this approach.

PagerDuty has introduced PD Automation Runner in Early Access to extend automated incident response to on-premises and self-hosted environments, beginning with Kubernetes. Deployed within a customer’s infrastructure, the lightweight component initiates outbound connections to PagerDuty, allowing Workflow Actions and the SRE Agent to authenticate locally, gather diagnostics, execute approved remediations, and return results without requiring inbound ports, IP allow lists, or persistent external access. The company describes a workflow in which incidents can trigger automated pod analysis, log review, historical incident comparison, and recommended fixes that responders can approve for local execution. PagerDuty says the runner also captures diagnostic and remediation data to improve future SRE Agent recommendations across cloud and on-premises systems. Support for additional self-hosted tools, including Splunk, Grafana, and Prometheus, is planned.

PagerDuty has introduced Scoped OAuth for Public Apps in Early Access to give administrators centralized visibility and control over third-party applications accessing their accounts. The feature replaces broad, difficult-to-track API key access with explicit, resource-specific permissions, allowing apps to request only capabilities such as reading incidents rather than full account access. Admins can review an app’s publisher and requested scopes before installation, approve or deny access, see authorized users, revoke individual or all connections, and fully uninstall apps when needed. PagerDuty positions the capability as a foundation for secure autonomous operations, where integrations, workflows, and AI agents increasingly act on data at scale and require transparent, approved, and rapidly revocable access.