The AI agent now understands the images a customer sends, not only the text they type. A screenshot of an error, a photo of a product, a picture of an item that arrived damaged: the agent reads what is in the image and uses it in the same conversation. It is live now, with nothing to switch on.
What changes inside a conversation
Until now, when a customer pasted a screenshot or snapped a photo, the agent saw an attachment it could not open. The conversation stalled until a person looked at it. Now the agent reads the image itself and answers when the answer depends on what the picture shows. When it does need a human, it hands the chat over with the image already understood, so the agent picking it up starts with context instead of from scratch.
The same applies to custom skills. A skill can branch on what is in an image, so an automated flow reacts to what the customer actually showed, not only to what they typed.
The buying signals hiding in images
A shared image is often intent in disguise. A shopper sends a photo of something they saw elsewhere and asks whether you carry it. Someone close to buying sends a screenshot of their cart with a question about shipping. A customer uploads a picture of the room they want to furnish. Each of those is a buying signal, and until now it was a signal the automated side of the conversation could not read.
The agent can now pull those signals out of the image, qualify them, and keep the conversation moving toward an order or a captured lead. For an ecommerce team, that means more high-intent conversations the agent can carry on its own and fewer that go quiet because nobody was free to open the attachment.
Available now
Live across chats handled by the AI agent. Send it an image in a test conversation to see how it reads it.
Get a summary with