The non-text side of WhatsApp: how an AI agent reads emoji and uses them in replies, what it does with stickers, GIFs and reactions, how voice notes are handled, and how to set emoji rules for your brand

Real WhatsApp is not paragraphs. It is a thumbs-up, a sticker, a forty-second voice note, a location pin and a ๐Ÿ™. An agent that only understands typed sentences replies to a thumbs-up with a paragraph, and that is how customers learn it is a bot in the bad way. Here is how each kind of message should be handled.

By Ritchie, ReplyKit ยท Published 5 October 2026 ยท 6 min read

Quick answer

An AI agent reads emoji as part of the message (a ๐Ÿ™ after a confirmation is thanks, a ๐Ÿ˜ก is a tone signal), uses emoji in replies only within the rules you set (typically at most one, matching the customer's style), acknowledges stickers and GIFs briefly or not at all rather than treating them as questions, does not reply to reactions, transcribes voice notes and answers them like text, and handles locations and forwarded messages as information. You set the brand line in the persona: no emoji for a law firm, one for a cafe. ReplyKit handles each type this way by default.

Message type by message typeVoice notesEmoji rules for your brandTestingFAQ

Message type by message type

MessageWhat the agent does
Emoji in textReads it as tone and meaning; ๐Ÿ‘ after an offer is a yes, ๐Ÿ˜ก raises the care level
Emoji only (๐Ÿ‘, ๐Ÿ™, โค๏ธ)Treats as acknowledgement; replies briefly or not at all depending on context
Sticker or GIFAcknowledges lightly if a reply is needed; never interprets it as a request
Reaction to a messageNo reply
Voice noteTranscribed and answered as text; the agent replies in text
Location pinUsed as the customer's location for service-area and directions questions
Forwarded messageRead as information; the agent asks what the customer needs with it
Photo or documentDescribed and used; see the photo article below

Photos are covered in how agents understand photos, and typed informality in slang, typos and mixed languages.

Voice notes

Voice notes are a large share of messages in many markets. The agent transcribes the audio, reads the transcript like a text message, and answers in text. Accents and background noise reduce accuracy; if the transcript is unclear the agent asks a short clarifying question rather than guessing. For businesses whose customers mostly send voice notes (trades, older customers, drivers), tell the agent in its instructions to confirm the key details it heard before acting on them.

Emoji rules for your brand

Conservative (law, finance, medical)

Do not use emoji. Read the customer's emoji for tone but reply in plain text.

Neutral (most services and shops)

Use at most one emoji per message, only when the customer uses them, and never in confirmations of dates, prices or policies.

Warm (cafes, salons, lifestyle)

You may use one or two friendly emoji in greetings and thanks, matching the customer's style; none in prices or policies.

Stickers

Do not send stickers or GIFs.

The reason to keep emoji out of confirmations is clarity: a date with a ๐ŸŽ‰ is still clear; a price with a ๐Ÿ˜‰ looks negotiable. The persona work is in persona and tone of voice.

Testing

Send the agent a thumbs-up, a sticker, a voice note, a location and a ๐Ÿ™ in the sandbox, and check it does not write a paragraph for any of them. Then send "ok ๐Ÿ‘" after a booking offer and check it books. The routine is in testing before going live.

Frequently asked questions

Can the agent understand voice notes?

Yes. It transcribes them and answers in text, asking for clarification if the audio is unclear.

What does it do with a sticker?

Acknowledges briefly if a reply is needed, or nothing. It never treats a sticker as a request.

Does it reply to reactions?

No.

Will it use emoji?

Only within the rules you set: none, one, or a couple, and never in prices or policies.

Does a thumbs-up count as a yes?

In context, yes; after an offered slot the agent books it.

Understands the thumbs-up and the voice note

Every WhatsApp message type handled, emoji within your brand rules. 7-day free trial, no card.

Start free
Written by Ritchie, ReplyKitPart of the small team in Malaysia that builds and runs ReplyKit. Writes about WhatsApp automation, AI customer service and small business lead response.

How we know this. ReplyKit agents receive every WhatsApp message type; the behaviours here are the defaults and the rules customers most often adjust. Product figures in this article (limits, prices, per-reply costs) are taken from ReplyKit as it runs today and are re-checked when we update the page. Where we cite outside research, the source is linked below. We sell ReplyKit, so read our product claims with that in mind; we say where a different tool or no tool is the better choice. About ReplyKit.

Sources and further reading

  1. Harvard Business Review, The Short Life of Online Sales Leads (Oldroyd, McElheran, Elkington, 2011) ยท firms that responded within an hour were about seven times more likely to qualify the lead
  2. WhatsApp Business Messaging Policy ยท opt-in, opt-out and prohibited-use rules every business must follow

ReplyKit is an independent product and is not affiliated with or endorsed by Meta or WhatsApp. This article is general guidance for running a small business, not legal, financial or regulatory advice.