Using AI to pre-mortem an incident runbook, Part 1
Have AI walk through an incident runbook step by step and flag failure modes before a real outage.
35 min · Reviewed 2026
The premise
Runbooks rot quietly until the next incident exposes the gaps. AI can pre-mortem a runbook step by step and surface the failure modes you missed.
What AI does well here
Walk through each step and propose what could go wrong.
Spot steps that assume access or context not stated.
Suggest verification points after each action.
What AI cannot do
Test the runbook against your real systems.
Know which engineer hates which tool.
Replace a chaos engineering drill.
Using AI to draft a vendor renewal questionnaire
The premise
Vendor renewals get rubber-stamped because nobody asks the owner the right questions in time. AI can draft a tight questionnaire that forces real answers.
What AI does well here
Draft 10 to 12 questions covering usage, satisfaction, and overlap.
Tailor questions to the vendor category.
Suggest a usage-data screenshot to attach.
What AI cannot do
Pull real usage data from the vendor.
Negotiate the renewal price.
Know the politics around the original purchase.
Using AI to design a shift handoff template
The premise
Shift handoff templates either drown the next shift in detail or leave critical actions buried. AI can design a template that separates know-this from do-this.
What AI does well here
Design a template with explicit do-this and know-this sections.
Suggest fields for ongoing incidents and watch items.
Recommend a max length per section.
What AI cannot do
Know which on-call always over-shares.
Enforce that people actually use the template.
Replace a verbal handoff for complex incidents.
Using AI to draft a supplier risk tier rubric
The premise
Without a tiering rubric, every supplier is treated the same and important ones get under-reviewed. AI can draft a defensible rubric in under an hour.
What AI does well here
Define 3 to 4 tiers with criteria.
Map diligence steps required per tier.
Suggest review cadence per tier.
What AI cannot do
Tier specific suppliers without seeing your data.
Substitute for procurement and legal review.
Know which supplier the CEO has a relationship with.
Using AI to inventory process debt
The premise
Process debt is hard to see because everyone has adapted around it. AI can read across a quarter of meeting notes and tickets and name the recurring workarounds.
What AI does well here
Spot recurring workarounds across notes.
Group by process area.
Suggest the underlying broken assumption per cluster.
What AI cannot do
Know the politics of fixing each one.
Quantify the time cost without instrumentation.
Tell you which to fix first based on strategy.
Using AI to draft warehouse slotting rationale
The premise
Slotting plans are usually a spreadsheet with no narrative. AI can draft the rationale so the plan can be reviewed, questioned, and revisited.
What AI does well here
Translate a slotting spreadsheet into a written rationale.
Highlight the top velocity assumptions driving the plan.
Suggest review triggers.
What AI cannot do
Walk the warehouse and check ergonomics.
Know the picker tribal knowledge about specific SKUs.
Account for seasonal pairings not in the data.
Using AI to audit customer service macros
The premise
Macro libraries grow until nobody knows what is in them. AI can audit the library for stale references, overlap, and tone drift in a single pass.
What AI does well here
Spot macros referencing deprecated features.
Cluster overlapping macros.
Flag tone outliers against a brand guide.
What AI cannot do
Know which macros agents actually use.
Test rewrites against real customer satisfaction.
Decide which deprecated features still need a graceful response.
Using AI to design a quality control checklist
The premise
QC checklists tend to grow until inspectors skip steps. AI can design a tighter checklist focused on the failures that actually escape.
What AI does well here
Prioritize checks by escape frequency.
Suggest leading indicators tied to past escapes.
Recommend which checks to drop.
What AI cannot do
Know operator skill differences across shifts.
Test the new checklist on the line.
Decide regulatory minimums you cannot drop.
Using AI to rewrite a 30-60-90 onboarding plan
The premise
Generic 30-60-90 plans give new hires comfort but not direction. AI can rewrite one into role-specific outcomes a manager can grade.
What AI does well here
Convert vague goals into measurable outcomes.
Suggest stakeholder shadowing slots.
Add explicit owner-of-help per outcome.
What AI cannot do
Know your team's actual bandwidth to support new hires.
Pick the right buddy for the new hire's personality.
Replace a real first-week conversation with the manager.
Using AI to recommend dashboard metric decommissions
The premise
Dashboards collect metrics nobody uses or trusts. AI can review them and recommend retirements, merges, and renames before the next quarterly review.
What AI does well here
Spot duplicative metrics across dashboards.
Flag metrics with confusing names.
Suggest retirements based on usage data when provided.
What AI cannot do
Know which metric a board member quotes once a year.
Decide policy on metric ownership.
Migrate alerts and runbooks that depend on the metric.
AI Supplier Risk Radar: Spotting Vendor Red Flags Before They Hit Operations
The premise
AI can dramatically speed up supplier risk radar by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for supplier risk radar from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on supplier risk radar — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Warehouse Slotting Rationale: Pick-Path Logic You Can Defend
The premise
AI can dramatically speed up warehouse slotting rationale by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for warehouse slotting rationale from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on warehouse slotting rationale — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Shift Bid Conflict Detector: Surfacing Coverage Gaps Early
The premise
AI can dramatically speed up shift bid conflict detector by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for shift bid conflict detector from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on shift bid conflict detector — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Incident Postmortem Scaffolder: Turning Pages of Logs into Timelines
The premise
AI can dramatically speed up incident postmortem scaffolder by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for incident postmortem scaffolder from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on incident postmortem scaffolder — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI SOP Drift Monitor: Catching When Practice Walks Away From Procedure
The premise
AI can dramatically speed up sop drift monitor by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for sop drift monitor from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on sop drift monitor — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Capacity Planning Narratives: Translating Forecast Models for Leaders
The premise
AI can dramatically speed up capacity planning narratives by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for capacity planning narratives from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on capacity planning narratives — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Third-Party Audit Prep: Evidence Mapping Without Last-Minute Panic
The premise
AI can dramatically speed up third-party audit prep by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for third-party audit prep from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on third-party audit prep — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Procurement RFP Summarizer: Apples-to-Apples Comparisons in Hours
The premise
AI can dramatically speed up procurement rfp summarizer by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for procurement rfp summarizer from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on procurement rfp summarizer — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Changeover Checklist Builder: Production Line Switchovers Done Right, Part 2
The premise
AI can dramatically speed up changeover checklist builder by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for changeover checklist builder from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on changeover checklist builder — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Quality Defect Cluster Notes: Patterns Across a Week of Returns
The premise
AI can dramatically speed up quality defect cluster notes by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for quality defect cluster notes from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on quality defect cluster notes — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Fleet Maintenance Triage: Sorting Tickets by Risk and Cost
The premise
AI can dramatically speed up fleet maintenance triage by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for fleet maintenance triage from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on fleet maintenance triage — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Vendor Onboarding Runbook: From Contract to First Delivery
The premise
AI can dramatically speed up vendor onboarding runbook by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for vendor onboarding runbook from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on vendor onboarding runbook — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Warehouse Safety Walk Summaries: Consistent Hazard Reports
The premise
AI can dramatically speed up warehouse safety walk summaries by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for warehouse safety walk summaries from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on warehouse safety walk summaries — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Staffing Coverage Forecast Explainer: Why Tuesday Looks Tight
The premise
AI can dramatically speed up staffing coverage forecast explainer by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for staffing coverage forecast explainer from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on staffing coverage forecast explainer — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Customer Escalation Classifier: Routing Tickets to the Right Owner
The premise
AI can dramatically speed up customer escalation classifier by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for customer escalation classifier from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on customer escalation classifier — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Ops Meeting Action Extractor: Owners, Dates, Decisions
The premise
AI can dramatically speed up ops meeting action extractor by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for ops meeting action extractor from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on ops meeting action extractor — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Route Exception Explainer: Why a Driver Deviated
The premise
AI can dramatically speed up route exception explainer by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for route exception explainer from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on route exception explainer — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Inventory Cycle Count Narratives: Variance Stories for Finance
The premise
AI can dramatically speed up inventory cycle count narratives by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for inventory cycle count narratives from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on inventory cycle count narratives — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Launch Readiness Status Rollup: Cross-Team Snapshot in Plain English
The premise
AI can dramatically speed up launch readiness status rollup by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for launch readiness status rollup from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on launch readiness status rollup — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Ops Dashboard Anomaly Callouts: Surfacing What Changed Overnight
The premise
AI can dramatically speed up ops dashboard anomaly callouts by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for ops dashboard anomaly callouts from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on ops dashboard anomaly callouts — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Business Continuity Tabletop Drafts: Realistic Disruption Scenarios
The premise
AI can dramatically speed up business continuity tabletop drafts by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for business continuity tabletop drafts from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on business continuity tabletop drafts — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Purchase Order Discrepancy Notes: PO vs Invoice in One Page
The premise
AI can dramatically speed up purchase order discrepancy notes by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for purchase order discrepancy notes from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on purchase order discrepancy notes — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Warehouse Training Microlearning Drafts: 5-Minute Skill Refreshers
The premise
AI can dramatically speed up warehouse training microlearning drafts by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for warehouse training microlearning drafts from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on warehouse training microlearning drafts — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Supplier Scorecard Narratives: Numbers Plus Context for Reviews
The premise
AI can dramatically speed up supplier scorecard narratives by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for supplier scorecard narratives from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on supplier scorecard narratives — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
AI Ops Runbook Rewrite Pass: From Tribal Knowledge to Shareable Doc
The premise
AI can dramatically speed up ops runbook rewrite pass by handling structure, summarization, and first-pass synthesis — but the professional judgment, verification, and final accountability stay firmly with you. This lesson shows where the leverage is real and where the guardrails belong.
What AI does well here
Produce a structured first draft for ops runbook rewrite pass from the inputs you provide.
Summarize long documents, threads, or logs into consistent formats your team already uses.
Compare items side by side against a checklist or reference you supply.
Reformat content for different audiences (executive, frontline, regulator, customer) without losing the underlying facts.
What AI cannot do
Make the final professional judgment call on ops runbook rewrite pass — accountability stays with you.
Confirm facts, citations, dates, dollar amounts, or legal authority — those must be verified against source records.
Replace the relationships, context, and institutional knowledge that make your decisions defensible.
End-of-lesson check
15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-operations-AI-and-incident-runbook-pre-mortem-r9a2-adults
What is the primary purpose of having AI conduct a pre-mortem on an incident runbook?
To document all past incidents for historical reference
To automatically fix any errors found in the runbook
To replace the need for on-call engineers entirely
To identify potential failure modes before an actual incident occurs
Which task is WITHIN AI's capability when pre-morteming a runbook?
Testing the runbook against production systems to verify it works
Replacing a chaos engineering drill entirely
Enumerating failure modes that could occur at each step
Predicting which engineer will be on-call during the next incident
Why should a team still conduct a game day (chaos engineering drill) after using AI for a runbook pre-mortem?
An AI pre-mortem complements but does not replace testing against real systems
Game days are the only way to update runbook documentation
Game days are required by compliance regulations
AI pre-mortems are too expensive to run frequently
What type of runbook step would AI most likely flag as a potential failure mode?
A step that includes timeout values for each action
A step that clearly explains how to restart a service
A step that assumes the engineer has access credentials that weren't provided
A step with a detailed rollback procedure
What should you ask AI to add to a runbook step if it's missing?
A humorous anecdote to keep engineers engaged
A verify-after step to confirm the action succeeded
A link to the team's favorite music playlist
A longer description using more technical jargon
Which of the following represents a realistic LIMITATION of using AI for runbook pre-mortems?
AI cannot parse text-based runbooks
AI cannot identify steps that assume unstated context
AI will always produce accurate runbooks on the first attempt
AI cannot know which engineer prefers which troubleshooting tool
A runbook step says 'Reset the database connection pool' but doesn't specify which system credentials to use. What problem would an AI pre-mortem likely identify?
The step lacks sufficient context about required access or credentials
The step needs more emojis to be engaging
The step uses too many technical terms
The step should be written in a different programming language
After pasting a runbook into an AI, which action aligns with the lesson's recommended workflow?
Ask the AI to schedule a game day for the team
Ask the AI to write a new runbook from scratch
Ask the AI to immediately implement all suggested changes
Ask the AI to enumerate failure modes, flag access issues, and suggest verify steps
An AI pre-mortem identifies that a critical runbook step has no verification step. Why is this problematic during an actual incident?
Verification steps slow down incident response unnecessarily
Verification steps are only needed for non-critical steps
Without verification, an engineer might proceed assuming the action succeeded when it actually failed
AI cannot actually identify missing verification steps
Why is it valuable for AI to analyze a runbook step-by-step rather than as a whole document?
Step-by-step analysis uses less AI computing resources
Step-by-step analysis is not actually more effective
It allows AI to identify failure modes specific to each action and where context is assumed
AI can only analyze small amounts of text at once
A team uses AI to pre-mortem their runbook and receives a list of potential failures. What's the next logical step?
Review and incorporate the findings, then plan a game day to validate
Delete the runbook and start fresh based on AI output
Immediately update the runbook with all AI suggestions without review
Ignore the findings since AI might be wrong
What does the lesson mean when it says runbooks 'rot quietly'?
Runbook problems only become visible when tested under real incident conditions
Runbooks are being actively attacked by malware
Runbooks are literally decaying due to hardware failure
Runbooks generate error logs automatically
Which statement best captures the value an AI brings to runbook review that a human might miss?
AI can remember every runbook ever written by the team
AI can automatically deploy fixes to production systems
AI can determine which team member wrote each step
AI can systematically analyze every step without fatigue and spot assumed context humans overlook
A runbook contains a step: 'Update the DNS cache on the load balancer.' The AI flags this as potentially problematic. What is the most likely reason?
The step should use IP addresses instead of DNS
The AI cannot analyze DNS-related steps
The step assumes the engineer has access to the load balancer without stating it
DNS is not a valid technical term
Why can't an AI pre-mortem completely replace chaos engineering or game day exercises?
Game days are too expensive to run
Game days are not allowed in most organizations
AI and game days serve the same purpose
AI can only analyze theoretical scenarios, not actual system behavior under failure conditions
