This essay is based on a conversation with Sean McDonnell, 43, who lives in the UK. Mr. McDonnell is the founder of web design company Kaizen and SaaS website Consigns. The following has been edited for length and clarity.
Last month, hours after I started trying out Anthropic’s new Fable model, I was halfway through it when the U.S. government forced Anthropic to cut off international access to Fable 5 with little notice.
When Fable ended, I was well prepared and had a backup plan in place, so it didn’t have a hugely disruptive impact on my web design business and the SaaS website I started earlier this year.
We are a small team, so AI tools are very helpful. Without AI, websites and other services could not be as advanced as they are. But this incident served as a reminder that a Plan B, or even a Plan C, is absolutely necessary when dealing with AI.
The mistake many people make is relying entirely on AI.
When Fable came back, we were already preparing for it to disappear again
General access to Fable has been restored and you are already using it. The first thing I did was make sure everything in the entire codebase was saved as a Claude skill.
That way, even if your model becomes inaccessible again, other models can refer to that skill and get a good map of your codebase from a more advanced model.
It’s important to keep records that exist outside of your AI tools and ensure they are documented during development. If Claude knows everything about our codebase and it’s being pulled tomorrow, can he pass the details to the developer? I think he can at this stage since we’re documenting everything that’s going on. It’s failsafe.
The key is to enhance, rather than replace, normal workflows with AI.
When Fable went down for the first time, I had another backup plan in place
Previously, when using Claude’s Opus 4.6, it would quickly hit the token limit and stop mid-task. We didn’t realize how token-intensive this tool was, and it left our codebase in a bit of a mess. Having learned this lesson in 4.6, I was better prepared for unexpected situations when I started using Fable for the first time last month.
Last month, before foreign access was cut off, I asked Fabre to create a guide that both Claude and other AI models could follow. This allowed remaining tasks to be handed off to other agents even if Fable became inaccessible.
I passed some to Codex and others to Claude 4.8. Without such preparation, a lot of work could have been wasted due to fable problems.
Read more from founders primarily working on AI
I hope there will be better communication from the company if something like this happens again
More notifications would be very helpful, but I can also understand why there were no notifications if there was a “security” issue. However, companies that rely on AI, including Anthropic, OpenAI, Gemini, other AI companies, and small businesses like us, should always have backup plans and prepare for disruption.
I would have liked a smoother transition instead of simply yanking the model offline.
Messages that explain the situation within your own app are fine.
Do you have a similar story? If so, please contact our reporter. aapplegate@businessinsider.com.