The customer wants to speak to a person: what your ChatME assistant actually does
A customer asks to speak to a person in the middle of a conversation with your assistant. What really happens at that moment, what the assistant never promises, and how the request reaches you to answer yourself.
ChatME··5 min read
"Can I speak to someone?" The message lands in the middle of a conversation with your assistant, and the first reaction isn't about the customer — it's about you. What if the bot just loops them around? What if it answers something odd and the customer only realises there's no one behind it once they're already annoyed? Some businesses hold off on turning on WhatsApp or Instagram for exactly this reason — not general distrust of AI, but the image of a real customer stuck talking to a machine that won't let them out.
It's worth saying plainly what actually happens at that moment, because the answer isn't "the assistant steps aside and a person appears" — that isn't what ChatME offers, and it isn't what any chatbot like this should promise today.
What really happens when someone asks
The assistant never pretends to be a person at any point: on its first reply in every conversation it identifies itself as an AI assistant. The EU AI Act has required this since August 2026, and ChatME does it on its own, with nothing for you to configure, on your website chat as well as on WhatsApp, Instagram, Messenger and Telegram. So when someone asks to speak to a person, they aren't discovering they were talking to a bot — they already knew.
What the assistant does next is collect the request, not pretend to resolve it itself. If you have a form with keywords like "speak to someone" or similar, the assistant detects it and starts asking for the details — name, phone, what they need — announcing it first with something like "I'm going to ask you a few quick questions." That request reaches you two ways at once: by email, to the address you've set under Forms, and saved in your dashboard under Forms → Submissions. The full conversation, with everything the customer wrote, stays in Conversations, inside your dashboard — not in your inbox. And this is the part no chatbot like this should hide: it's a request that you confirm and answer yourself. There's no "transfer to a human agent" button and no queue. The value the assistant adds at that exact moment is that nobody is left hanging at 11pm on a Sunday waiting for you to open — it captures the contact right there, and the gap between that message and your reply becomes as short as you can make it, not as long as it takes you to next check your phone.
If you're already answering from your own phone
This is different from someone explicitly asking to speak to a person, and it's worth not mixing the two up: if you step into that WhatsApp or Instagram conversation yourself, from your own phone, without the customer asking for anything, the assistant detects it and steps back from that specific conversation for a while — 30 minutes by default, adjustable from zero minutes up to a full day depending on the channel — so the two of you don't answer the same person at once. It keeps answering every other conversation normally; this doesn't turn the assistant off, it only pauses it on that one thread while you're handling it. Once the time passes it comes back, and since the customer just spoke with a real person, it identifies itself as AI again on its next message — it never stops saying so, not even right after you've stepped in.
It's built around the real case, not the ideal one: you have time to read a message and reply yourself sometimes, not always, and when you can't, someone needs to be there. On WhatsApp this works as soon as you connect your number with the "Keep your phone" option; on Instagram it's the same behaviour, with the minutes configurable from that section of your dashboard. On Messenger, today, the assistant doesn't detect your manual replies from the Facebook app, so the automatic pause doesn't exist there yet — a real difference between channels that's worth knowing before it catches you off guard with a customer.
What the assistant can't do, and it's better to know upfront
It doesn't book a free table you never told it existed, it doesn't invent a price when a quote depends on seeing the job, and it doesn't decide anything that needs human judgement — the EU AI Act directly bans using a system like this for high-risk decisions, and ChatME wouldn't allow it either way. What it does do is answer with the real information you've given it, capture the request when it can't go further, and leave you the part that genuinely needs a person. That's less flashy than promising an "always-available human agent," but it's what actually happens when someone hits send.
For a small business with no dedicated front desk — a one-chair hair salon, a two-person accounting firm — this is usually a better deal than pretending to be bigger than you are: the customer knows what to expect, and you're not promising call-centre availability you don't have. If instead you get a lot of these requests every day and genuinely need someone available in real time, the assistant can keep capturing everything while you decide whether it's worth having someone watching the dashboard during those hours.
If you want to see what a "speak to someone" form would look like on your own assistant, or you're unsure which keywords should trigger it — and only then — write to us at info@chatme.es. And if you don't have WhatsApp or Instagram connected yet, you can see what each plan includes at pricing.
*Availability may vary by country.*
START TODAY
Ten minutes. Try it free.
Try it on your real website. If you’re not convinced, cancel without writing a single email.