Customer Sends a Photo, the Bot Identifies the Product

TL;DR: A customer sends a photo in chat and aiSTAFF identifies the product or answers about it, then matches it to your catalog with price and availability, on any channel, from the same shared brain.

People send photos, not product codes

Watch how a real customer asks for something. They rarely know the model name or the SKU. They snap a photo of a chair they saw at a friend's place, a screenshot of an item from an ad, or a picture of the broken part they need to replace, and they send it with "do you have this?" A text-only bot is stuck there. It cannot see the photo, so it asks the customer to describe the thing they already showed it, and the conversation stalls.

aiSTAFF reads the image. The customer sends the picture, and the bot identifies the product or answers the question about it, then moves straight into matching it against your catalog. The friction of describing a visual thing in words disappears. If you want this on your channels, the AI agents service sets it up, and it is one feature of the wider aiSTAFF platform.

What vision does in a chat

Image understanding is not a gimmick bolted on the side. It plugs into the same selling flow as a text question:

• Identify the product. The bot recognizes what is in the photo and names it.
• Match the catalog. It searches your embedded catalog for that product or the closest equivalent, returning a card with price, discount, and availability.
• Answer about it. Questions like "is this waterproof" or "what sizes does this come in" get answered from your product data.
• Continue the sale. It suggests related items, checks stock before confirming, and offers a callback if intent is high.

So a photo is a starting point for the same conversation a typed query would start, with the same guardrails. The catalog match runs on the hybrid search engine described in the platform hub, and the relevance gate still applies: if the photographed item is nothing like your stock, the bot says so instead of inventing a match.

It works on every channel

Vision is not limited to one surface. Customers send photos most on WhatsApp and Instagram, where snapping and sending is second nature, and aiSTAFF handles the image on both. Because the feature reads from the one shared brain, the same vision behavior shows up wherever a customer can attach an image. Putting an agent on WhatsApp, where photo messages are most common, is covered in an AI agent on WhatsApp Business, and the single-brain design behind it is in one brain across five channels.

A photo plus a Georgian question

The image and the text work together. A customer can send a photo with a caption in Georgian, and the bot reads both: it identifies the product from the picture and answers the caption in Georgian. It detects the language from the text and replies in kind, so a Russian caption gets a Russian answer. The catalog itself can be in English while the customer shops in Georgian, which is the cross-language behavior explained in making a chatbot speak fluent Georgian. The customer never has to translate their own request.

A worked example

A plumbing-supply store gets a WhatsApp message at 8pm: a photo of a corroded faucet cartridge and the caption, in Georgian, "I need this part, do you have it?" A human clerk would squint at the photo, try to match it from memory, and probably ask the customer to come in. aiSTAFF identifies the cartridge type from the image, searches the catalog, finds two compatible parts, and replies with both, showing price and stock, in Georgian. It asks whether the customer wants one set aside for pickup and captures a phone number. The store opens the next day to a ready order tied to a specific part, from a photo that would have stumped a text bot. The customer solved a visual problem with a picture, the way they wanted to.

That is the pattern: the customer does the easy thing, send a photo, and the bot does the hard thing, identify and match. It keeps the conversation moving instead of bouncing it back to the customer, which is the same persona discipline described in the chatbot that does not sound like a bot.

Honest limits

Vision is strong at identifying a clear product photo and matching it to your catalog. A blurry, dark, or ambiguous image can be hard to place, and in those cases the bot asks a clarifying question rather than guessing wrong. It also still respects the relevance gate, so a photo of something you do not sell returns an honest "we do not carry that" instead of a forced match. And as everywhere in aiSTAFF, the outcome is product discovery plus a lead or callback, not an in-chat card payment. How much prior context the bot keeps while it works through a photo conversation depends on your plan, covered in conversation memory tiers, and tone per channel is set in per-channel tone control.

Related reading

• aiSTAFF: One AI Brain Across Every Channel
• Put an AI Agent on WhatsApp Business
• One AI Brain, Five Channels
• The Chatbot That Does Not Sound Like a Bot

FAQ

Can the bot really read a photo a customer sends?

Yes. aiSTAFF reads the image, identifies the product or answers the question about it, then searches your catalog for a match with price and availability.

Which channels support photo messages?

Vision works wherever a customer can attach an image, including WhatsApp and Instagram where photo messages are most common, all from the same shared brain.

What if the photo is unclear?

If an image is blurry or ambiguous, the bot asks a clarifying question instead of guessing. A photo of something you do not stock returns an honest no match.

Can a customer send a photo with a Georgian caption?

Yes. The bot reads the image and the caption together, identifies the product, and answers the caption in the customer language, Georgian, Russian, or English.