Skip to main content
AI generating a first draft of a post-mortem, populating each section from incident data Gathering the context to write a post-mortem is time-consuming. Information has to be collected from Slack threads, monitoring tools, investigation findings, and conversations with the people involved. Then it all has to be condensed into something coherent and readable. AI-native writing takes care of the heavy lifting. It can generate a first draft from your incident data, review your finished document with inline suggestions, and help you redraft sections that need improvement. You focus on the analysis and the learnings, and let AI handle the gathering and structuring. AI features are available on Pro and Enterprise plans.
For private incidents, AI features are only available if your organization has opted in to sending private incident data to AI subprocessors. See Private incidents.

First draft

Getting started

To enable AI-generated post-mortems, you need a template with AI enabled on at least one section. The quickest way:
  1. Go to Settings > Post-mortems and click Add template.
  2. Select Duplicate existing template with AI to create a copy of one of your existing templates with AI enabled on all custom sections. Or pick one of our suggested templates, which come with AI pre-configured.
  3. Optionally, open the template and customize the AI instructions on each section to guide what the AI focuses on.
You can also enable AI on an existing template by editing it and toggling AI on for individual sections. Enabling AI drafting on a template section with custom instructions Once your template is ready, create a post-mortem from an incident using that template and the AI will generate a first draft for you.

How it works

The AI generates content for each section of your template individually, using everything it knows about the incident: the timeline, Slack or Teams conversations, investigation findings, custom fields, catalog data, images, and more. The quality of the output depends on how you’ve configured your template. Each section can have its own AI instructions that guide what the AI focuses on. For example, you might tell it to focus on customer impact in one section and technical root cause in another. The section name and help text also feed into the generation, so being specific about what you’re looking for in each section makes a real difference. If a section doesn’t have AI enabled, it’s left with the template prefill content for the writer to fill in manually. This means you can mix AI-generated and manually-written sections in the same document — use AI for the sections where gathering context is the bottleneck, and leave the more reflective sections for humans. The result is a starting point, not a finished document. It gets you past the blank page and gives you something to react to and refine, rather than having to write everything from scratch. You can edit any of the generated content, and use Redraft to have AI rework specific passages.

Review

Once you’ve written your post-mortem (or refined the AI-generated draft), you can ask AI to review it. Click the Review button in the document header and AI will read through your document, compare it against the incident data, and leave inline suggestions. The review checks for things like:
  • Factual accuracy: are there events missing from the timeline, or details that don’t match what actually happened?
  • Completeness: are there contributing factors or context that the document doesn’t cover?
  • Structure and clarity: is the narrative easy to follow? Is it blameless in tone?
  • Learning value: are the lessons specific and actionable? Are the follow-ups concrete enough to actually get done?
Suggestions show up as highlighted annotations in the document, visible only to you. You can work through them one by one, or dismiss all suggestions at once. Leaving the document clears them as well — they’re meant to be acted on in the moment, not left as permanent comments.

Redraft

Sometimes you don’t need a full review, you just need help with a specific passage. Select any text in the document and you can ask AI to rework it. You tell it what you want (rewrite this more clearly, fact-check this against the incident data, make this less technical, change the tone) and it gives you a suggested replacement. This is useful for polishing sections after you’ve done the thinking but the writing isn’t quite there yet. The AI has access to the full incident context, so it can fact-check claims against what actually happened and suggest corrections. On top of that, it can find and update anything you reference, replacing plain text with rich elements such as user mentions, PR references, incidents, custom fields, and more.

Chat

The AI chat is a floating assistant available while you’re writing. You can ask it questions about the incident (“what happened between 2am and 3am?”, “who was on call when this started?”, “what did the investigation find?”) and it will answer based on the incident data. You can also use it to help with writing. Ask it to draft a paragraph about the customer impact, or to summarize the resolution steps. If the conversation surfaces action items, the AI can create follow-ups directly from the chat. When the AI references existing text in your document, you can click the reference to scroll straight to that part of the document.