What Happens When Your AI Receptionist Doesn't Know the Answer?
This is usually the real question behind "can I trust an AI to answer my customers" -- not whether it sounds smart, but what it does the moment it doesn't actually know something.
A well-built AI receptionist doesn't guess. Botnira only answers from the business description and FAQs it's actually been trained on -- if a question falls outside that, it falls back to a message you've written in advance, which typically transfers the conversation to your team or logs it for follow-up. That's a deliberate safety feature, not a gap in capability.
The real fear isn't "sounds robotic" -- it's "makes something up"
Most business owners evaluating an AI receptionist aren't primarily worried about the voice or the wording. They're worried about a customer asking something specific -- a price, a policy, whether a service is actually offered -- and the AI confidently giving a wrong answer. That's a legitimate concern with generative AI in general, and it's exactly what a good fallback design is built to prevent.
How Botnira is built to never guess
Botnira answers strictly from the business description and FAQs you write in Bot Studio -- it isn't pulling general knowledge about "what businesses like yours usually do" to fill gaps. When a question falls outside what you've actually written, it falls back to a fallback message you set up in advance, rather than improvising. That fallback typically transfers the conversation to a human or logs the message for your team to follow up on.
That means how complete your business description is directly determines how many conversations Botnira can resolve on its own -- a thin business description leads to more fallbacks, and a thorough one leads to fewer. It's a feature, not a limitation: the fallback rate is actually a useful signal for what to add to your training next.
Where the conversation goes next
Straight to your unified inbox
Every conversation Botnira hands off lands in your unified inbox, shown as a written thread regardless of which channel it came in on -- voice, WhatsApp, SMS, email or web chat.
A human takes over from there
Any team member with inbox access can jump into that specific conversation and reply directly, picking up exactly where the AI receptionist left off, on the same channel the customer used.
Common questions
Will an AI receptionist make up an answer if it doesn't know something?
A well-built one shouldn't. Botnira is designed to answer only from the business description and FAQs it's been trained on -- if a question falls outside that, it falls back to a fallback message you write, rather than guessing.
What is a fallback message?
It's a message you set up in Bot Studio that Botnira uses whenever a question falls outside your business description and FAQs -- typically transferring the conversation to a human or logging it for follow-up, instead of improvising a response.
Who handles a conversation after it's handed off?
Your own team, through Botnira's unified inbox -- a human agent can see the full conversation and jump in and reply directly, on whichever channel the customer originally reached out on.
Does a more complete business description reduce hand-offs?
Yes. How thorough your business description and FAQs are directly determines how many conversations your AI receptionist can resolve on its own -- gaps in training show up as fallback conversations, which is also a useful signal for what to add next.
Related reading
Training Your AI Receptionist
Where the business description, FAQs and fallback message are set.
Read the guide →Writing FAQs That Prevent Wrong Answers
When to use an FAQ instead of the business description, and how to write ones that hold up.
Read the guide →Using the Unified Inbox
How a human takes over a conversation Botnira has escalated.
Read the guide →