website_bot

A chatbot to connect to my personal website and explain my research.

Features

• Fetches and provides access to research papers and leaderboard data.
• Supports interactive chat functionality with history.
• Built using FastAPI for the backend and integrates with Pydantic AI.

Installation

Clone the repository:

git clone https://github.com/grgkovac/website_bot.git
cd website_bot

Create conda env:

conda create --name website_bot_env python=3.10
conda activate website_bot_env

Install dependencies:
pip install -r requirements.txt

Setup logfire:

logfire auth # follow the instructions to authenticate
logfire projects new # create a new project and note the project ID
# or
logfire projects use <project_id> # use an existing project

Run the application:
python main.py

Test the chatbot using the dummy client:
python dummy_client.py

API Endpoints

POST /chat

• Description: Handles chat requests and returns the agent's response.
• Request Body:

{
  "message": "Your message here",
  "history": []
}

Response:

{
  "reply": "Agent's reply",
  "new_history": []
}

Deploy with Google Cloud Run

Google cloud

Project setup:

gcloud auth login
gcloud config set project <your-project-id>
gcloud services enable run.googleapis.com cloudbuild.googleapis.com secretmanager.googleapis.com

Store Gemini API Key

echo -n "YOUR_API_KEY" | gcloud secrets create GOOGLE_API_KEY --data-file=-

Store OpenAI API Key

echo -n "YOUR_API_KEY" | gcloud secrets create OPENAI_API_KEY --data-file=-

Store Logfire Token

echo -n "YOUR_LOGFIRE_TOKEN" | gcloud secrets create LOGFIRE_TOKEN --data-file=-

Grant Permission: Allow the Cloud Run Service Account to read these secrets:

gcloud secrets add-iam-policy-binding GOOGLE_API_KEY \
    --member="<your-service-account>" \
    --role="roles/secretmanager.secretAccessor"

gcloud secrets add-iam-policy-binding OPENAI_API_KEY \
    --member="<your-service-account>" \
    --role="roles/secretmanager.secretAccessor"

gcloud secrets add-iam-policy-binding LOGFIRE_TOKEN \
    --member="<your-service-account>" \
    --role="roles/secretmanager.secretAccessor"

Deploy to Cloud Run:

gcloud run deploy websitebotapi \
  --source . \
  --project <your-project-id> \
  --region <your-region> \
  --allow-unauthenticated \
  --set-secrets="GOOGLE_API_KEY=GOOGLE_API_KEY:latest,LOGFIRE_TOKEN=LOGFIRE_TOKEN:latest,OPENAI_API_KEY=OPENAI_API_KEY:latest"

You can now access the deployed service. To test change the url in the dummy_client.py to the Cloud Run service URL and run: python dummy_client.py