Edition 4
The Disappearing Model
What happens when the AI your business depends on is no longer there?
From the Editor
“The interesting questions are often around the model.”
The most fragile AI dependency is often not an endpoint but a behaviour: the accumulated way a product, team and set of controls have learned to work with one model. This edition follows that dependency from first failure through evidence, economics, engineering and a practical test.
Dr Gary Marrs PhD Editor, Beyond the Model
AI models increasingly sit inside critical business processes. They are also products controlled by someone else.
This edition asks what happens when those two facts collide. It begins with a failure, then traces why a nominally compatible replacement may not be equivalent, what the published evidence says about model lifecycles, how to judge the cost of resilience, and how to build and rehearse a proportionate escape route.
The conclusion is not that every service needs several providers. AI can still be the best economic and technical choice. The question is whether an organisation understands the dependency it has created, and has deliberately chosen how long it can operate without that capability.
Inside Edition 4
In this edition
6 connected questions, designed to be read in sequence or entered at the point most relevant to you.
- 0101 — The failureThe Disappearing ModelYour AI provider is still online—but the model your business depends on has disappeared. What does it take to restore the business capability?
- 0202 — The lock-inSame API. Different Machine.Why changing a model name does not necessarily restore the behaviour, controls or service level an AI product depends on.
- 0303 — The evidenceModels Have Expiry DatesThe published lifecycle evidence shows why AI infrastructure can have a shorter and more provider-controlled life than it appears.
- 0404 — The economicsThe Price of an Escape RoutePortability has a real cost. This decision framework helps judge when the price of an AI escape route is worth paying.
- 0505 — The engineeringBuild for the Model You Haven't Chosen YetPractical engineering patterns for model substitution, including internal contracts, evaluation suites, shadow testing and a proportionate fallback path.
- 0606 — The testTomorrow, Your Model Is GoneA practical 30-day exercise for testing how an organisation would cope if its primary AI model became unavailable tomorrow.