A message blocked by mistake leaves your AI employee available
For everyone who works with an AI employee
Part of When an AI employee is unavailable, the right person is told
When the AI provider's automatic safety filter blocks one of your messages by mistake, your AI employee stays available: you get a short, plain explanation, and your next messages are answered without the blocked one.
Why it matters
An automatic filter can block a normal business message by mistake. One blocked message should not look like a broken AI employee, hold up the rest of the conversation or alert people who have nothing to fix.
- Before
- A few blocked messages in a row could put the AI employee in its unavailable mode, alert your admins and keep the blocked text in the way of the next answers.
- Now
- Messages blocked by the AI provider's safety filter get a plain explanation, alert nobody and stay out of what your AI employee reads for your next messages.
What you get
- A plain explanation: the reply says that the AI provider's automatic safety filter blocked this message, that such filters sometimes block normal messages by mistake, and that your access is fine.
- Your AI employee stays on: a blocked message never counts as a failure of the AI account, so nobody is alerted and everyone keeps getting answers.
- A clean next message: your AI employee does not read the blocked message or the explanation again, so sending it again or rephrasing it starts clean. If the same words are blocked again, rephrase them.
- Still in your chat: your message and the explanation stay in the chat for you to read and search.
- An empty reply is not a used-up plan: when the AI provider sends back no text, you are told that no reply was generated, and a single empty reply alerts nobody.
- Real problems still reach the right person: an expired sign-in, a used-up plan or repeated failures are reported to the person who can fix them, as before.
How to use it
Nothing to switch on: it applies to every chat with an AI employee, with your own Claude or ChatGPT plan and with VEP usage.
Part of: When an AI employee is unavailable, the right person is told
If an AI employee is unavailable, the person who can fix it gets one alert, others get a short plain note, and unanswered questions come back later.
More updates to this feature
- One plain note when your AI employee could not save what it learnedFixedFor everyone
- A reply that did not finish says so in one plain sentenceFixedFor everyone
- A failed chat command gives you a reference for VEP SupportFixedFor everyone
- Report a problem to VEP Support and it reaches a personImprovedFor everyone