A broken part, a damaged parcel, a model plate, the haircut they want. Customers send a photo, your AI assistant reads it and answers from your own content.
Half of support questions are hard to put into words. "The little plastic thing under the handle snapped." "The box arrived like this." "Can you do this colour?" A photo answers the question before the customer has finished typing it.
With photo questions on, a camera button appears in the message bar of your chat widget. Customers pick a photo or take one, add a line of text if they like, and send it. On WhatsApp, Instagram and Messenger they simply send the photo in the conversation.
Your assistant describes what is in the photo, looks that up in your knowledge base, and answers with the image in view. It copies any model or part number it can read, and when it cannot be sure, it says so and asks for a clearer photo or a photo of the model plate.
The description of the photo is used to search your knowledge base, so the answer comes from your manuals, price lists and policies, not from general knowledge.
Numbers and text visible in the photo are copied exactly. A brand, model or number it cannot see is reported as not visible instead of invented.
Location and other metadata are removed first. The photo is read once to answer, then discarded. Only a short text description stays in the conversation, under your normal logging and encryption settings.
Website widget, direct link and QR code, plus WhatsApp, Instagram and Messenger on Growth and Scale. Same assistant, same knowledge base.
An offensive or explicit photo gets a short refusal and counts toward the same abuse protection as abusive text, so it can end the chat.
Turn it on per chatbot with one switch. Bots that do not need photos never show the button.
See how solo professionals use this feature every day.
A customer photographs a worn pump fitting. The assistant reads the part number stamped on it, finds it in the uploaded parts list, and replies with the replacement and its price.
A parcel arrives crushed. The customer sends a photo, the assistant explains the damaged-delivery policy from your returns document and tells them exactly what happens next.
A client sends a photo of nail art from Instagram. The assistant says which service it is, what it costs and how long it takes, and offers a booking.
A guest sends a photo of the hot water panel. The assistant recognises it from the house manual and explains which switch to press.
In the editor, open the Window tab, scroll to Features and switch on File Upload. The camera button then appears in the message bar of your widget. On WhatsApp, Instagram and Messenger, photos are answered automatically once the channel is connected.
Any paid plan, with the BotChap AI assistant as your chatbot backend. WhatsApp, Instagram and Messenger photos need Growth or Scale, because those channels do.
Reading a photo is one extra AI request, billed the same way as a reply: from your AI Credits, or by your own provider if your website chatbot uses your own API key. WhatsApp, Instagram and Messenger always use AI Credits.
No. The photo is processed in memory to describe and answer it, then discarded. Location and camera metadata are removed before anything else happens. Only the short text description is saved with the conversation.
One clear photo of the thing in question, in good light, with any label or model plate visible. One photo per message. JPG, PNG and WebP photos are accepted.
It is not designed to. It answers questions about your business from your own content. If your customers might send photos of health or safety issues, add a line to your system prompt telling the assistant to recommend a professional.
Start free, no credit card needed. Upgrade to unlock AI features.