HeroVerge

HeroVerge

HeroVerge is a tool built for providing Gen-AI-based assistance to Support Engineers, maximizing productivity and managing extensive workloads.

For deploying in local server(flask version) pull the entire repository and run app.py.

• LLM API creds would be made available on request

Features

• Query Handling: Takes user input as a query and returns responses from a pretrained LLM of the selective choice. Currently available models are: llama3_70b, llama3_8b, and llama2.
• JSON Responses: Returns JSON responses with 5 keys: summary, script, sentiment, complete, and notes. These responses can be configured at the backend to retain ticket data and provide a guide for further resolution.
• Client Sharable Email: Frames a client-sharable email to address the query and take forward the next steps. LangChain is used to create a chain out of the JSON responses and cook up an email-based response from the JSON created by the AI model.

Technologies Used

Deployments

There are two deployments of this project across two Python frameworks:

Flask Version

• Runs on a local server
• Supports robust designs and functionality using HTML, CSS, and JS

Streamlit Version

• Deployed on Hugging Face
• Streamlit Deployment Link

Both versions have different UI and frontend configurations. Unlike Streamlit/Gradio, Flask supports HTML, CSS, and JS for more robust designs and functionality.

UI Appearance ?️

Flask Version (localhost)

Streamlit Version (Hugging Face)

Example Test Run Case ?

• Query Used: "Hi support, can you please help me fix my problems with your APIs?"
• Model Selected: llama3_70b

API Response

{
  "summary": "Customer is experiencing issues with the APIs",
  "script": "Can you please provide more details about the issues you're facing with our APIs?",
  "sentiment": 50,
  "complete": false,
  "notes": "Customer is experiencing unknown issues with the APIs, need more information to troubleshoot"
}

Future Development

Ticketing System: Facilitate recording and tracking of issues and grant the LLM access to this data to generate insights. Roles and Permissions: Allot admin and user roles so only admins can create insights from the ticketing database using credentials.

Contact ? For more information or to get in touch, please email [email protected] Phone number : 9179041912