Skip to main content

AI Auto-Responses

The core of Remo is its ability to hold natural, helpful conversations with your clients on WhatsApp — automatically, at any time of day or night.

How It Works​

When a client sends a message, the bot doesn't use a fixed script or keyword matching. Instead, it reads the conversation as a whole and generates a tailored response based on:

  • What the client just said
  • What was discussed in previous messages
  • The bot's instructions and personality settings
  • Your business's products, services, and policies

This means every response is relevant, coherent, and consistent with your brand — even across long, multi-turn conversations.

What the Bot Remembers​

Two things keep the bot on the thread of a long conversation:

  • The 30 most recent messages, word for word — everything recent is available exactly as it was written, so a preference the client mentioned twenty messages ago is still in play
  • A rolling summary of everything older — as a conversation grows past that window, the earlier part is condensed into a running summary rather than being forgotten

The practical effect is that the bot stops losing the plot in long chats. A client who comes back after a week of back-and-forth doesn't have to re-explain what they're looking for.

Customizing the Bot's Personality​

You control how the bot speaks to clients through the instructions you write in Configuration → Advanced, on the AI System Prompts card. You can instruct the bot to:

  • Use a formal or casual tone
  • Greet clients by name
  • Emphasize specific selling points or features
  • Avoid certain topics (e.g., "do not discuss pricing directly — direct clients to speak with an agent")
  • Always end with a specific call to action (e.g., "always offer to schedule a site visit")

Example:

"You are a warm and professional assistant for [Company Name]. Always address clients by their first name, focus on the lifestyle benefits of our development, and invite every interested client to schedule a visit."

The better your instructions, the more effectively the bot represents your business.

Write the Prompt You'd Give a New Hire

Short, vague prompts produce short, vague bots. Facts that change often — hours, phases, current promotions — are easier to maintain as Project characteristics, which the bot reads as background context on every reply.

What the Bot Can Answer​

The bot is capable of addressing a wide range of client questions, including:

  • Availability and details of your products or properties
  • Appointment scheduling and available time slots
  • Your office locations and how to get there
  • Documents, brochures, and images (sent directly in the chat)
  • General business information — hours, team, processes

Which of these the bot actually offers depends on what you've enabled and what content exists behind it. See Routes & Capabilities.

Voice Messages​

If a client sends a voice message, the bot automatically:

  1. Downloads and transcribes the audio
  2. Reads the transcription to understand what the client said
  3. Replies appropriately, just as it would to a text message

The transcription is written into the conversation in your dashboard, so your team can read what was said without playing the audio back.

The bot can also reply with a voice note of its own when the client wrote by voice — a client who spoke gets spoken back to. Voice replies aren't on by default and consume extra credits. See Voice Notes & AI Phone Calls for what's required and what it costs.

Emoji Reactions​

A client reacting to one of the bot's messages with an emoji is treated as a real turn in the conversation, not as noise. The bot decides whether the reaction deserves an answer:

  • A simple acknowledgement stays unanswered — a 👍 on a message the bot just sent doesn't need a reply, and the bot doesn't manufacture one
  • A reaction that reads as a question or a problem may get a response — something like a ❓ on a price or a plan is treated as the client asking about it

The bot never narrates the reaction back to the client. It either answers what the reaction implies, or stays quiet.

Multi-Message Responses​

For detailed or complex questions, the bot can split its answer across two or more consecutive messages. This keeps responses digestible and avoids large walls of text that feel unnatural in a messaging app.

When the Bot Doesn't Know the Answer​

If a client asks something the bot doesn't have information about, it responds gracefully — acknowledging the question and offering an alternative, such as speaking with a human agent or visiting in person. The bot is designed never to make up information it doesn't have.