🦙📚 TrevorText - Copilot-augmented

Empower Trevor Project counselors with RAG-assisted TrevorText, an AI co-pilot powered by LlamaIndex to help counselors focus on people, not paperwork. Link to Hackathon Submission: https://devpost.com/software/counselor-copilot

In this version, we artificially simulate the contact's responses. The extension would be to integrate the tool with the existing SalesForce CRM and webchat platform used by the Trevor Project.

Local Dev Loop

pyenv local Use python version in .python-version file

virtualenv -v venv-trevorhack-streamlit && source venv-trevorhack-streamlit/bin/activate Create and activate virtual environment

mkdir .streamlit && cd .streamlit && touch secrets.toml Create secrets file

openai_key = "XXXX" insert your API key for open ai / others like so

Create environment variable file with: touch .env and add the following variables (Reference)

ASTRA_DB_APPLICATION_TOKEN="TOKEN"
ASTRA_DB_API_ENDPOINT="API_ENDPOINT"
OPENAI_API_KEY="API_KEY"

pip install -r requirements.txt Install deps

streamlit run streamlit_app.py Run app from the project root

The dev server can be accessed at localhost:8501

Additional Notes:

kill -9 $(lsof -i:8501 -t) 2> /dev/null To quickly kill any process running on port 8501

Overview of the App

  • Takes user queries via Streamlit's st.chat_input and displays both user queries and model responses with st.chat_message
  • Uses LlamaIndex to load and index data and create a chat engine that will retrieve context from that data to respond to each user query
  • Copilot Agent takes actions using provided tools to search the web for resources, escalate cases, and retrieve relevant documents to inform recommended responses.

Demo App

TBC

Get an OpenAI API key

You can get your own OpenAI API key by following the following instructions:

  1. Go to https://platform.openai.com/account/api-keys.
  2. Click on the + Create new secret key button.
  3. Next, enter an identifier name (optional) and click on the Create secret key button.

Try it out

Once the app is loaded, enter your OpenAI API key (please forgive the ghost elements) and check out the suggested responses from the agent.

Edit data/library/demo_conversation_client.txt to try out responses to different statements.