firstbatchxyz / dria-searching-agent

Dria Search Agent

Dria Search Agent replies queries with scientifically grounded answers with a multi-agent system.

DRIA Search Agent

DRIA Searching Agent is an AI-powered tool that answers your questions by selecting the most relevant agent from a pool of agents with different backstories. The selected agent searches the internet, reads articles and websites, and provides a detailed response to your query.

Prerequisites

• Python 3.12 or 3.13
• Poetry
• Docker

Installation

1. Install Poetry:

   curl -sSL https://install.python-poetry.org | python3 -

2. Clone the repository:

   git clone [e-mail]:firstbatchxyz/dria-searching-agent.git
   cd dria-searching-agent

3. Install dependencies using Poetry:

   poetry install

Running the Application with Poetry

To run the app with poetry, first start the Qdrand and Browserless services using docker-compose

docker-compose up -d

Then run the application using Poetry in one of three modes:

poetry run search
poetry run search_v2 # with manager
poetry run server # server mode

Which compiles three different versions of the program.

This command sets up a virtual environment specific to the project and executes the app script or module specified in the pyproject.toml under [tool.poetry.scripts].

Running the Application in server mode with Docker-compose

Project can be run in a server mode with the following docker compose profile

docker-compose --profile server up

Which sets up Qdrant, Browserless and agent services

After the services are ready, agent server will be listening on port 5000, example request body:

{
    "query":"How does Google Maps detects traffic?",
    "with_manager": true
}

How to Use

1. Ask a question or provide a query.
2. The DRIA Searching Agent will select the most relevant agent from its pool based on the question.
3. The selected agent will search the internet, read articles, and gather information to provide a detailed response.
4. The agent will present the answer to your question, along with the sources used to compile the response.

Additional Tools and Commands

• Run tests with Poetry using a command like poetry run pytest.
• Update Python dependencies using poetry update.