Crimson Agent

Crimson Agent is a python-based framework built that can be used to build custom chatbots for privacy critical applications such as for banking and finance.

How to test it out?

Note: Before starting make sure you have python3 and pip installed on your system

1. Clone this repo:

git clone https://github.com/shahiryar/crimson-agent.git

2. Install dependencies

pip install -r requirements.txt

3. Run the app

streamlit run app.py

---

TODOs

Develop POC

• Demo Intent Classifier Training
• Intent Classification
• Follow up entity reprompt
• Sentiment Analysis

Make Robust

• Add No-Match Intent (threshold based)
• Redirect conversation amdist of active context (Cancel Slot Filling)
• Reponse formating from context
• Design Context Lifecycle Management
• Handle Input and Output contexts for Intents
• Modulate the code for reusablity

Make Configurable

• Configurations Page to add new Intents and Entities and a Configuration Page to add new agents
• Train and Retrain Script
• Session Lifecycle Management

Make Integrated

• Communication Channels Integration
• Enable Whatsapp Message Sending
• Enable Webhook Integrations
• Plan Deployment

Make Better

• Option for Rule-based Intent Matching
• Implement open-source LLM to generate Natural Language Responses
• Manage Knowledge Base Creation and Connection