A7medAyman/Agentic-Procurement-Assistant

Agentic Procurement Assistant

A modular, agent-based system that automates product research, extraction, and procurement report generation using AI and web automation.

Table of Contents

1. 1.Overview
2. 2.Architecture & Workflow
3. 3.Workflow Diagram
4. 4.File Structure Diagram
5. 5.Technologies Used
6. 6.Detailed File Explanations
7. 7.Dockerfile
8. 8.main.py
9. 9.unit_test/
10. 10.routers/
11. 11.helpers/
12. 12.agents/
13. 13.controllers/
14. 14.Agentic Procurement Assistant API
15. 15.Endpoint
16. 16.Input Parameters
17. 17.Response
18. 18.Quickstart
19. 19.Contact

Overview

Agentic Procurement Assistant leverages a crew of AI agents to automate the procurement process:

• —Recommends search queries for products.
• —Searches e-commerce platforms.
• —Scrapes product details.
• —Generates a professional procurement report (HTML).

Architecture & Workflow

Workflow Diagram

flowchart TD
    A[User API Request] --> B[Search Queries Recommendation Agent]
    B --> C[Search Engine Agent]
    C --> D[Web Scraping Agent]
    D --> E[Procurement Report Author Agent]
    E --> F[HTML/JSON Output]

File Structure Diagram

graph TD
    A[src]
    A1[main.py] --> A
    A2[routers/] --> A
    A3[controllers/] --> A
    A4[agents/] --> A
    A5[helpers/] --> A
    A6[agent_output/] --> A
    A7[unit_test/] --> A
    B[notebooks/]
    C[Dockerfile]
    D[.env]
    E[requirements.txt]

Technologies Used

• —crewai
• —tavily-python
• —scrapegraph-py
• —fastapi
• —uvicorn[standard]
• —python-multipart
• —pydantic
• —python-dotenv
• —pytest

Detailed File Explanations

Dockerfile

• —Purpose: Containerizes the application for consistent deployment and execution.
• —What it does:
• —Sets up a Python environment.
• —Installs dependencies from requirements.txt.
• —Copies the application code into the container.
• —Sets environment variables and exposes the necessary port.
• —Defines the default command to run the FastAPI server.
• —Tools Used: Docker, Python, Uvicorn.

main.py

• —Purpose: Entry point for the FastAPI application.
• —What it does:
• —Initializes the FastAPI app.
• —Includes routers (API endpoints).
• —Starts the application server.
• —Tools Used: FastAPI, Uvicorn.

unit_test/

• —Purpose: Contains automated tests to ensure code correctness.
• —What it does:
• —test_procurement_report.py: Tests the procurement report generation logic, including agent orchestration.
• —Tools Used: Pytest, FastAPI TestClient.

routers/

• —Purpose: Defines API endpoints and request/response handling.
• —What it does:
• —CrewRouter.py: Implements the /api/get_procurement_review endpoint.
• —Calls the controller to run the agent workflow, and returns the output file.
• —Tools Used: FastAPI APIRouter, Pydantic.

helpers/

• —Purpose: Provides utility functions and configuration management.
• —What it does:
• —config.py: Loads environment variables and configuration settings (e.g., API keys, file paths) for use throughout the app.
• —Tools Used: Python os, dotenv.

agents/

• —Purpose: Defines the agents, their tasks, tools, and output schemas.
• —What it does:
• —config/agents.yaml: YAML file specifying agent role, goal, and backstory.
• —config/tasks.yaml: YAML file defining the sequence and dependencies of tasks.
• —output_schema.py: Pydantic models that define the structure of agent outputs (e.g., search results, product details).
• —tools.py: Integrates external tools (Tavily for search, Scrapegraph for scraping) and provides a unified interface for agents to use them.
• —Tools Used: CrewAI, Pydantic, Tavily, Scrapegraph, YAML.

controllers/

• —Purpose: Orchestrates the workflow of agents and manages the overall process.
• —What it does:
• —CrewControllers.py: Loads agent and task configurations, initializes agents, runs them in the correct order, collects outputs, and generates the final procurement report (HTML/JSON).
• —Tools Used: CrewAI, Python, file I/O.

Agentic Procurement Assistant API

Hosted at:

https://A7medAyman-Agentic-Procurement-Assistant.hf.space/api/get_procurement_review

Endpoint

Method: POST URL: /api/get_procurement_review Content-Type: multipart/form-data

Input Parameters

Response

• —File: procurement_report.html
• —Content: A professional HTML report with product tables, price comparisons, and recommendations.
• —test output

[image]

**note:** if you get this message as a response

  {"message": "An error occurred while running the crew litellm.APIConnectionError: Cohere_chatException - {\"id\":\"5cb1c64b-e3a1-4ff9-a06c-b8f436733d25\",\"message\":\"You are using a Trial key, which is limited to 10 API calls / minute. You can continue to use the Trial key for free or upgrade to a Production key with higher rate limits at 'https://dashboard.cohere.com/api-keys'. Contact us on 'https://discord.gg/XW44jPfYJu' or email us at support@cohere.com with any questions\"}."

This indicates that the free trial key from Cohere has hit its usage limit :(. ---

Quickstart

1. 1.Clone the repository:

   git clone git@github.com:AhmedAyman77/Agentic-Procurement-Assistant.git
   cd Agentic-Procurement-Assistant

1. 1.active your conda env

    conda activate your_env_name

1. 1.Install dependencies:

   pip install -r requirements.txt

1. 1.Set up environment variables:

   cp src/.env.example src/.env

Edit src/.env and fill in your API keys.

1. 1.Run the FastAPI server:

   cd src
   uvicorn main:app --reload --host 0.0.0.0 --port 5000

1. 1.Access the API docs:

   $BROWSER http://localhost:5000/docs

Contact

For questions or support, please contact: Email: devahmedaymn@gmail.com LinkIn: Link