Senior Solutions Consultant

Investigate smarter from any agent: blast radius and AI post-incident reviews come to the PagerDuty MCP Server. PagerDuty's new intelligent Model Context Protocol (MCP) tools, recently announced in Early Access, are a step toward autonomous operations. Here's how they help customers get there. The problem isn't your tools. It's the tabs between them. Incident response means a lot of tab switching for most teams. Every tab switch is a delay to resolving the issue... The information needed to resolve the incident exists. It's just scattered across surfaces that don't talk to each other, and the responder becomes the integration layer. AI agents are supposed to fix this, but an agent is only as good as the tools it's given. Early MCP implementations, started as plumbing: API endpoints converted to tools and handed to agents as-is. As the protocol has matured, so has what teams present to agents. At PagerDuty PagerDuty, Inc. think of MCP as a product surface, one that speaks to external agents and LLMs on behalf of your operations data, and it deserves the same design rigor as any UI. Tools should be shaped so agents act quickly, consume tokens efficiently, and return repeatable answers across the full range of real operations work: investigation, response, management, analytics, onboarding, reporting, and change management. What PagerDuty, Inc. is announcing. Using PagerDuty's Remote MCP Server, connected agents can run the full incident lifecycle: create, acknowledge, annotate, reassign, and resolve incidents, and look up who's on call, from supported MCP clients. What's new is Early Access to two intelligent tools that go beyond operations into investigation and learning: blast radius analysis and AI-drafted post-incident reviews. You can see these tools in action at Dreamforce in partnership with the Slackbot MCP Client, where a responder can run an entire incident, including accessing key data from PagerDuty, without leaving Slackbot. Let's look at what each tool does. Map the blast radius before your customers do. When a service degrades, the immediate question is scope... Answering that usually requires an expert that might not be available. The blast radius tool answers it from data. Built on PagerDuty's historical record of which services have had incidents together, it: * Maps the dependencies of the affected service * Identifies which downstream services are at risk * Surfaces correlation strength for each * Quantifies user and account impact. The responder can then ask their agent to check whether any of those correlated services have active incidents right now, which is often how the actual root cause surfaces. This analysis is possible because it draws on PagerDuty's historical incident graph. Draft the post-incident review without leaving the channel. The learning step is the one most teams skip, because by the time the incident resolves, everyone wants to go back to their day. The post-incident review tool helps make sure that step actually happens, right from your MCP client or terminal. A responder can ask their agent to draft a post-incident review for a resolved incident, and PagerDuty generates the AI draft with the timeline, involved services, participants, and narrative markers pre-populated from the incident record. The agent tracks generation progress, and once the draft is ready, the responder can ask for it and get a summary back in the channel, no trip into the PagerDuty UI required. The review gets written while the context is fresh, in the same place the incident was worked, saving time and helping ensure the context is documented for next time. Bringing it together. Its MCP covers the day-to-day mechanics of incident management. These two new tools add to what makes PagerDuty different: intelligence derived from years of your incident history, exposed where agents can act on it. Together they make PagerDuty a platform an agent can genuinely operate: the system of record, the historical intelligence layer, and the action surface, all through one well-designed MCP interface. Your team's surface of choice might be Slackbot today and something else next year. The intelligence underneath shouldn't care. None of this removes humans from the loop. The agent gathers context, surfaces analysis, and executes the actions a responder asks for. Judgment about what to do stays with the people accountable for the outcome, who now spend their attention on the decision instead of the data gathering. Paving the path to autonomous operations. Your team shouldn't be the integration layer between the systems that hold your incident context. PagerDuty's approach to autonomous operations: * Puts intelligent agents to work at scale, handling the noise, accelerating resolution, and keeping you in the loop where it matters * Deepens the full incident management lifecycle by empowering teams to resolve incidents faster * Broadens the platform and ecosystem with capabilities that help teams prevent incidents from happening The intelligent MCP tools move PagerDuty, Inc. closer to that vision. Ready to see how they can bring PagerDuty's intelligence to your team's surface of choice? The blast radius and post-incident review tools are available now in Early Access. Reach out to PagerDuty customer support to request access to the preview tool group for your account.

Control, AI, and ease | color commentary, aug 2026. Thu Sep 10 2026 Even though the end of summer is flying by, Statsig, Inc. still shipped a lot of new features in August. I want to check in on both what Statsig, Inc. shipped and why Statsig, Inc. shipped it. Its main goal in August was to continue to make Statsig work better for you, however you want to use it. The three big themes within that goal were scaling cross-platform control, building out more AI capabilities, and tackling customer ease-of-use requests. Scaling cross-platform control. Whether you use Statsig via the console, the API, the CLI, or with integrated tools, Statsig, Inc. want to make sure you can trust your experiment results and release metrics. "Have it your way" isn't just the slogan for my third-favorite burger joint. * PagerDuty integration lets Statsig Topline Alerts page PagerDuty directly to detect problems faster * API: Pull daily experiment checks for better visibility and delete single feature gate user overrides without having to mutate the full list locally * SCIM Role Aliases can now match custom IdP naming conventions without having to rename anything in your IdP * Warehouse-native: Cost attribution tags on Statsig queries help you better evaluate compute costs, and Athena projects now get full pipeline visibility out of the box AI mode. Statsig, Inc. believe that AI implemented correctly can help teams ship faster, and building out Statsig's AI capabilities is a top priority. With MCP use steadily increasing, it's clear you think this is a priority too. August saw five new fields added to MCP experiment configuration. If you want to create or update an experiment from your coding agent, you can now modify secondaryIDType, identifierMappingMode, enabledNonProdEnvironments, links, and secondaryMetrics, no console required. Ease of use: You asked, Statsig, Inc. built it. You're the experts at using Statsig, and you know (often better than Statsig, Inc.) what would make it easier to use. That's why Statsig, Inc. prioritize improvements from your requests. This month, Statsig, Inc. focused on requests for Experiment and Feature Gate creation modals. * Layers are now a top-level targeting option for experiments, making it easier to keep tests mutually exclusive and prevent cross-contamination * Targeting multiselect is expanded to better manage as many targeting values as you need * Display names can be shared across gates, experiments, and configs, so you don't run into "ID already in use" as often As always, if you have other requests, let Statsig, Inc. know! The next episode. It's been a packed summer expanding the foundational pillars of its platform and building out feature requests. Looking forward, Statsig, Inc. has got even more to come. Statsig, Inc. has a couple of big releases in the next two months, and Statsig, Inc. is making continued progress toward a combined, AI-powered behavioral and validation stack with Amplitude and Statsig. Statsig, Inc. is excited to show you what's up soon.