Platform · Core concepts

The nouns of MessageMind.

Account, AI model, contacts, channels, integrations, automations. Learn the mental model that every other page in this documentation hangs off.

  • One account, one AI model
  • Many channels, many integrations
  • Entity ledger dedups every event
Try MessageMind free Jump to reference

What's inside

The headlines before you dive in.

  • Account: the MessageMind tenant where every message, contact, conversation, automation and integration lives. One account connects one AI model, many channels and many integrations.

  • AI model: the brain of the account, trained on your catalogue, policies and tone, that answers every incoming message on every connected channel. One account has one active AI model.

  • AI model category 'eCommerce for Physical Goods': the model shape for merchants selling stocked SKUs, with product buckets for simple and variable products that inventory integrations (Shopify, Cin7 Core, Unleashed, Salesforce, AutoTrader) write into.

  • AI model category 'Services': the model shape for appointment-based businesses, with a services catalogue (name, duration, base price, category) that booking integrations (SimplyBook, Jobber, Housecall Pro) write into.

  • AI model category 'Hotels': the model shape for hospitality, with rooms, rates and stays that PMS integrations (Hostaway, Lodgify, Uplisting, Guesty) write into and that drives booking-request vs booking distinctions in the ledger.

  • Contact: a person the account has met, keyed by phone, email and per-channel identity fields (WhatsApp number, Instagram handle, Facebook page scoped id, Viber id, integration ids), deduplicated by an identity lock so one real person stays one contact across channels.

Documentation Platform fundamentals Core concepts

Platform integration

Core concepts

  • CategoryPlatform

Overview

MessageMind is a tenant-shaped platform that models your business as one account, one AI model that knows your catalogue and tone, and the people who message you as contacts across many channels. Every WhatsApp, SMS, call, email, Instagram, Facebook, Viber, TikTok and website-chatbot message that lands is resolved to a contact, attached to a conversation on the channel it arrived on, and answered by the account's AI model 24/7. Integrations plug third-party services (CRMs, PMSs, inventory, forms, calendars) into the same account so the AI can greet returning customers, read live stock, book appointments, or push outcomes back to the system of record; automations watch for triggers coming from channels and integrations and enrol contacts into campaigns. A per-account entity ledger stamps every webhook delivery, event and trigger firing, so a retried webhook, a parallel create or a sync replay is recognised as the same event and never doubles up a reply or a campaign.

What MessageMind can do with it

  • Account: the MessageMind tenant where every message, contact, conversation, automation and integration lives. One account connects one AI model, many channels and many integrations.
  • AI model: the brain of the account, trained on your catalogue, policies and tone, that answers every incoming message on every connected channel. One account has one active AI model.
  • AI model category 'eCommerce for Physical Goods': the model shape for merchants selling stocked SKUs, with product buckets for simple and variable products that inventory integrations (Shopify, Cin7 Core, Unleashed, Salesforce, AutoTrader) write into.
  • AI model category 'Services': the model shape for appointment-based businesses, with a services catalogue (name, duration, base price, category) that booking integrations (SimplyBook, Jobber, Housecall Pro) write into.
  • AI model category 'Hotels': the model shape for hospitality, with rooms, rates and stays that PMS integrations (Hostaway, Lodgify, Uplisting, Guesty) write into and that drives booking-request vs booking distinctions in the ledger.
  • Contact: a person the account has met, keyed by phone, email and per-channel identity fields (WhatsApp number, Instagram handle, Facebook page scoped id, Viber id, integration ids), deduplicated by an identity lock so one real person stays one contact across channels.
  • Conversation (thread): the running history of messages exchanged with one contact on one channel. A contact can hold many conversations (one per channel they reach you on) and each conversation keeps its own channel-specific state and transcript.
  • Channel: an inbound and outbound message surface the account owns, where real conversations happen. The current set is WhatsApp, SMS, Phone, Email, Instagram, Facebook, Viber, TikTok and the Website chatbot.
  • Integration: a third-party service plugged into the account to lend data or receive outcomes (HubSpot, Salesforce, Shopify, Odoo, Jobber, DenGro, Perspective, Google Sheets, Hostaway, and the rest). An integration is not a channel: it never carries a customer conversation on its own.
  • Automation: a trigger-based campaign that watches for a specific event (a new HubSpot contact, a Shopify order, a Jobber client, a sheet row, an inbound message) and enrols the matching contact into a sequence of steps on a chosen channel.
  • Entity ledger: the per-account record of every webhook delivery, external event and trigger firing, kept for 90 days. It is how MessageMind recognises a HubSpot retry, a parallel Jobber CLIENT_CREATE + REQUEST_CREATE, a ScoreApp redelivery or a Shopify replay as the same event, so a contact is never enrolled twice and a reply never fires twice.

AI agent use cases

  • The account is the container; everything else (AI model, contacts, conversations, channels, integrations, automations, ledger) hangs off it, and nothing crosses accounts.
  • Channels bring in messages from real people; integrations bring in data and events from other software. The AI model answers across every channel; integrations feed it context (a sales order to quote, a stock level to check, a room to offer) and receive outcomes (a contact update, a new row, a tag).
  • Every inbound message is resolved to a contact, attached to a conversation on its channel, and answered by the one AI model configured for the account's category (Physical Goods, Services or Hotels).
  • Automations sit on top: a trigger fires (from a channel or an integration), the entity ledger confirms it is not a duplicate, and the matching contact is enrolled into the campaign exactly once.

Example workflows

What happens when a message arrives

  1. A customer sends a message on a channel (WhatsApp, SMS, Phone, Email, Instagram, Facebook, Viber, TikTok or the Website chatbot). The channel's inbound webhook posts the payload to MessageMind.
  2. The entity ledger stamps the delivery against the account; a byte-identical retry from the channel is recognised and dropped so nothing fires twice.
  3. The sender is resolved to a contact by phone, email or per-channel identity (WhatsApp number, Instagram handle, Facebook page scoped id, Viber id). A new contact is created only when no identity matches; otherwise the existing contact is reused and the identity lock keeps one real person as one contact across channels.
  4. The message is appended to the conversation for that contact on that channel. If no conversation exists yet on that channel, one is opened.
  5. The account's AI model reads the conversation history plus any context its category gives it (catalogue for Physical Goods, services for Services, rooms and rates for Hotels, plus live reads from connected integrations such as a Shopify stock level or an Odoo sales order) and replies on the same channel.
  6. If the message matches an automation trigger (inbound keyword, first-message, channel-specific), the ledger confirms the contact has not already been enrolled on that trigger and the automation enrols the contact exactly once; its first step runs on the configured channel.