Deep dive · module 91 of 100
Fallback strategies
Ensure there is an alternative if AI is unavailable or fails.
After this lesson
You can describe a simple fallback for a AI-dependent task.
In view · Choices and control
A second route is ready
The AI step delivers a usable result within the agreed boundaries.
Recognise a malfunction, time-out, or unusable answer.
Route to a human, use a simple rule, or visibly postpone the task.
Part 1 of 3 · Normal route
Your preference for less movement is on. Choose each part yourself.
What is a fallback?
An alternative route if the AI service does not work, is too slow, or provides an unusable answer.
Examples
A human queue, a simpler rule-based version, or postponing the task.
Test the fallback
Check whether the alternative route also works and is clear to users.
New practice example
From explanation to a usable result.
A fallback route must state what is known, what failed, and how someone can safely proceed.
Fictional practice situationA fictitious assistant cannot retrieve the confirmed reservation conditions. The user asks if a booking can be cancelled free of charge.
View a possible solution
This is how it can be done
Report that the source is currently unavailable and do not invent a cancellation period. Refer to the person in charge or a confirmed alternative source. Keep the question as 'to be processed' without stating that no conditions exist.
Why this works
No access to evidence does not mean there is no evidence or rule. A fallback answer must keep that distinction visible.
Make it slightly more difficult
Describe the fallback route for a calculation error, an unreachable source, and an uncertain sending status. Which cases are allowed to automatically retry?
Extra practice, at your own pace. You do not need to submit anything. This does not count towards an automatic assessment.
Your workplace · learning by doing
Make something. See what changes.
Checking the AI connection… You can already view the written example.
Present your assignment to Ende.
Read the answer, change one thing, and compare.
Previously held
New version
You remain in control
What do you want to improve in your next attempt?
On this device only. AI answers remain only in this page memory. This is self-review, not a grade.
Why this approach works
No access to evidence does not mean there is no evidence or rule. A fallback answer must keep that distinction visible.
A step further
Describe the fallback route for a calculation error, an unreachable source, and an uncertain sending status. Which cases are allowed to automatically retry?
Try & discover · no obligation
Control an agent with a control port
Try to send before the check. Then also test a missing source or a malfunction.
- Retrieve source
- Check draft
- Execute action
What is happening?
- No source has been retrieved yet. Feel free to try the submit button first.
All tools, queues and submissions have been simulated. No message is sent and no external service is called.
This trial works locally with fictitious data and does not count towards your lesson completion. It starts anew with a new lesson or reload.
A prompt to get you started
Describe a fallback for a AI assistant that answers customer questions. What happens if the AI is unavailable?
Apply it
A customer assistant does not respond within the agreed waiting time. The original question is still available.
Which approach best fits this situation?
- Keep trying indefinitely without notification.
- Provide a clear notification and transfer the question to the agreed human queue.
- Make up an answer to hide the waiting time.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
Why is a fallback important?
- So that the process can continue if AI fails.
- It is not necessary.
- Only for large systems.
Extra practice
Choose a AI application and note a fallback that you would implement.
If the AI does not respond within 10 seconds, the question goes to an employee.
