This week I sat with three of the world’s largest financial institutions. Household names. Every one of them said the same thing: theatre is over.
Your IT DR plan should run, not be read
Most enterprise DR sits in a system somewhere, neatly catalogued, audited, signed off. Looks like resilience. Reads like resilience. But when the outage hits, it can’t be run. It’s maybe at most referenced, then abandoned.
The only IT DR capability that matters: in a real incident, does your recovery capability execute, or get thrown aside while the team improvises?
The test is simple. Can your DR investment actually cut recovery time in a real outage? If it can’t, you didn’t buy resilience. You bought a binder.
Most DR plans fail years before the crisis
Most disaster recovery plans don’t fail in a crisis. They fail years before it, in a spreadsheet nobody had time to fix. Fragmented runbooks. Hundreds of hours spent manually consolidating data. Governance that exists on paper but nobody enforces. Every migration starting from zero.
There’s a faster way in. It’s not a new platform. Teams are turning messy, unstructured DR plans into standards-compliant runbooks using tools they already have:
- Claude, with two skills added
- Your own master templates
- A connection to your Cutover instance via our open-source MCP server
Here’s the part that matters. It doesn’t write runbooks freehand. It reads each source plan, matches it to the right master template in your instance, and builds from there. Structure and governance are baked in before the AI does anything. When a source plan is vague or doesn’t map cleanly, it flags it. No guessing. Human stays in the loop.
No new vendor. No approval cycle. The AI tool your team already has, doing in minutes what used to take hundreds of hours.
Frequently asked questions
What is resilience theatre in IT disaster recovery?
DR that looks and reads like resilience, catalogued, audited and signed off, but cannot actually be run when an outage hits, so it is referenced at most and then abandoned.
Why do disaster recovery plans fail before a crisis?
They decay years earlier in spreadsheets nobody has time to fix: fragmented runbooks, manual data consolidation, unenforced governance, and every migration starting from zero.
How can teams turn existing DR plans into executable runbooks quickly?
By using an AI model they already have, with skills and their own master templates, connected to Cutover through the open-source MCP server. It matches each source plan to the right template and flags anything that does not map cleanly, with a human in the loop.