Maxbot

Maxbot is an open source library and framework for creating conversational apps.

Features (work in progress)

• Unified and extendable messaging over any platform.
• Integration with external libraries and NLU services. Support for active learning.
• DSL for declarative dialog description based on ready-made models of conversational logic.
• An engine for creating your own dialog models.
• NLG command language and templates.
• Implementing business logic in a general-purpose programming language.
• Tracking dialog. Sessions. Dialog state. Error handling and recovery.
• Complex bots with multiple skills. Interaction of skills.
• Developing, debugging, testing and logging tools.
• Deployment in any environment: stand-alone, cloud-enabled, hosted. Scaling.
• Integration with external data sources, services, information systems and support services.
• Business analytics. Insights.

Maxbot is the right balance of simplicity and flexibility based on time-tested solutions.

Documentation

Install Maxbot

For detailed installation instructions, see the documentation.

• Operating system: macOS / OS X · Unix/Linux · Windows
• Python version: Python 3.9-3.11 (only 64 bit)
• Package managers: pip

Pip

Using pip, Maxbot releases are available as source packages and binary wheels. Before you install Maxbot and its dependencies, make sure that your pip and wheel are up to date.

When using pip it is generally recommended to install packages in a virtual environment to avoid modifying system state:

python -m venv .env
source .env/bin/activate
pip install -U pip wheel
pip install maxbot

Quick Start

For detailed installation instructions, see the documentation.

First, we configure channels. Channels are a way to integrate your bot with various messaging platforms. You must configure at least one channel to create a bot. The telegram channel is configured by specifying secret api_token for the telegram bot.

Telegram channel is the best choice for a quick start because it's easy to run right on your laptop. Telegram Bot API allows you to receive incoming updates via long polling, so you don't need to have an external IP and set up webhooks.

Save the bot scenario as bot.yaml or something similar.

channels:
  telegram:
    api_token: 110201543:AAHdqTcvCH1vGWJxfSeofSAs0K5PALDsaw
intents:
  - name: greetings
    examples:
      - Good morning
      - Hello
      - Hi
  - name: ending
    examples:
      - Goodbye
      - Bye
      - See you
dialog:
  - condition: intents.greetings
    response: |
      Good day to you!
  - condition: intents.ending
    response: |
      OK. See you later.
  - condition: true
    response: |
      Sorry I don't understand.

Run the MaxBot CLI app passing the path to the bot.yaml as a parameter.

$ maxbot run --bot bot.yaml
✓ Started polling updater... Press 'Ctrl-C' to exit.

Open your bot in telegram messenger and greet it.

• Type Hello. You will see the appropriate response: "Good day to you."
• Type bye and the bot will reply "OK. See you later."
• Enter a message unknown to the bot, How are you? Get a response: "Sorry I don't understand."

The output in your console will look like this

[01/27/23 23:06:44], telegram#123456789
🧑 Hello
🤖 Good day to you!

[01/27/23 23:06:48], telegram#123456789
🧑 bye
🤖 OK. See you later.

[01/27/23 23:07:03], telegram#123456789
🧑 How are you?
🤖 Sorry I don't understand.

Press Ctrl-C to exit MaxBot CLI app.

Congratulations! You have successfully created and launched a simple bot and chatted with it.

Advanced examples

There are several examples of services built on Maxbot. They show the advanced features of Maxbot, such as custom messanger controls, integration with different REST services, databases and so on. You can also check the implementation details of these features in the examples below.

• Bank Bot example.
• Taxi Bot example.
• Transport Bot example.

Where to ask questions

The Maxbot project is maintained by the Maxbot team.

The near future

• Integration of deterministic dialog models (Dialog Tree, Slot Filling) with large language models (LLM) such as ChatGPT, LLaMA, etc.
• Improved debugging and logging tools.
• Released new examples of using the library to create complex bots.