Chatbot Concepts

Chatbots, or “bots” for short, are computer programs that interact with people in a way that mimics human interaction to some degree. The interaction can vary in complexity from simple keyword-driven queries to elaborate conversational systems using natural language processing and AI techniques. This conversational form of human-computer interaction can act as a framework for many types of useful applications.

Early chatbots included academic curiosities like Eliza (1964) and Julia (1994), but now bots are at work all over the internet in customer support, as virtual assistants like Google Assistant, and in many other roles.

Bots in Google Chat

Bots in Google Chat appear as special users marked BOT. Just like people, bots can participate in chat spaces, and they can respond to direct messages. But unlike people, bots only see messages directed to them when a person @mentions them in a chat space, or when a person sends them a direct message.

Bot access to user data

In order to operate in a useful way, a bot needs to know who is invoking it, and how to address that person. Beyond this basic identity data, bots do not have access to user data unless granted explicitly by the user:

  • By default, bots can only read the basic identity of users that invoke them. This information includes the user's display name, user ID, email address, and avatar image.
  • If a bot needs to access other data to do its job, it can prompt the user to grant it additional access.

Independence from users

Bots aren't associated with any particular user: anyone can add a bot to a space or remove it. Once the bot is added to a space, anyone can @mention it.

Some use cases for bots

Adding bots to a conversational platform like Google Chat lets people ask questions, issue commands, and so on without changing context. On its back end, a bot can access other systems, acting as an intermediary to those systems.

This ability to provide access to a wide range of resources and tools, while maintaining a unified cognitive experience, can provide an excellent framework for many types of applications, including:

  • workflow management
  • setup and configuration
  • order generation
  • search
  • data collection

Specific examples

Some low-level examples of things you might ask a bot to do from within the chat space include:

  • Look up information — A chatbot can retrieve information based on structured or free text queries entered by the user.
  • File ticket — A chatbot can generate a new incident report or other artifact, using information provided by the user.
  • Coordinate effort — A chatbot can enhance the way that team members interact with each other, by providing “team memory”, scheduling resources, and so on.

Integrating with Google Workspace

If you're considering building a Google Chat bot that will integrate with Google Workspace, consider implementing the bot in Apps Script. This provides your bot with easy access to Google Workspace services and data such as Sheets, Slides, Docs, and Calendar.

Google Chat API

The Google Chat API provides a number of ways that you can integrate your bots into chat spaces and DMs, but does not provide or dictate any particular way to implement the bot logic. You can create a simple command-driven bot or use any kind of language processing and AI services or modules that you like, and you can do this on a variety of platforms, including:

  • Apps Script
  • Google Cloud Platform
  • Cloud or on-prem HTTP server

See Bot implementation architectures for more on this topic.

Standard features

When you configure a bot in Google Chat, it must provide an identity; this consists of:

  • The bot name
  • An avatar image for the bot
  • Identity of the bot developer

Bots should also provide a help command or the equivalent to help users get started using the bot.

Bot communication with Google Chat

Bots can receive messages from, and send messages to, Google Chat. These messages use defined message formats, which may render as simple text messages or interactive cards in Google Chat.

Bots in direct messages (DM)

You can configure a bot so that it responds to direct messages. This provides a one-to-one conversational interface that involves a single user.

Bots in chat spaces

You can add a bot to a chat space, so that it can respond to specific content that appears in the space. For example, you can specify that any message that begins with "@foobot" is passed to some foobot app and the response appears in the chat space. The following sequence diagram summarizes the bot behavior:

Sequence diagram of chat bot in chat space

Google Chat communicates with your bot via an endpoint, which may be one of the following types:

  • HTTP (web service) endpoints can accommodate various styles of interaction.
  • Cloud Pub/Sub endpoints let your bot subscribe to a topic that relays messages from Google Chat, even if the implementation is behind a firewall.

See the Endpoint types section for more about the different ways you can connect Google Chat with your bot.