Blog: Resilience
If your AI stops, can your business keep its promises?
Published by Forever Foundry
An AI assistant becomes useful. People build it into their daily work. Eventually, a process depends on it.
Then access disappears.
The business question is how long customers can wait while you recover.
Continuity is becoming part of the AI conversation
On 20 October 2026, Microsoft announced new digital sovereignty capabilities and business continuity guidance. Its approach distinguishes between workloads that can use public cloud services and those needing greater operational independence.
The announcement includes a public preview of transitions from connected to disconnected operations. These are capabilities with different availability stages, not a guarantee that any organisation can switch to offline working without disruption. Read Microsoft’s announcement.
The FOFO™ view
Continuity requirements should be understood before an AI investment becomes an operational dependency.
Imagine a distributor using AI to help resolve order exceptions.
If it becomes unavailable, can staff still find the order history, understand previous decisions and authorise the next action? Is there enough trained capacity to handle the queue manually?
IT may know how to restore the service. Operations knows how quickly the backlog grows. Customer service knows which delivery promises will fail. Finance can assess the cost of delay.
Those perspectives belong in the same investment decision.
A fallback written in a document is only useful if people can perform it with the information and time available. The same is true when the missing resource is a person.
Price the fallback into the business case
Before making AI essential to a workflow, agree:
- Which customer commitments depend on it?
- How long can the business tolerate its absence?
- What minimum service must continue?
- Who can run the alternative process, using which information?
- How will work be reconciled when the system returns?
Test a realistic interruption. Measure the time to resume work, the backlog created, errors and recovery effort. Include the cost of maintaining the fallback when comparing delivery options.
The right response might be a manual procedure, better access to records, additional resilience or a different deployment model. It depends on the business consequence.
AI Compass™ listens across the organisation to surface constraints and dependencies. Slid3rs™ helps teams compare opportunities by business value and delivery complexity. FlowRoom™ supports agreement on priorities and ownership.
An AI investment should improve performance while preserving the organisation’s ability to serve its customers when conditions change.
Related reading
Next