readme.md

---

The CosmosAIGraph (caig) application is deployed as these two microservices:

Name	Functionality
web	Web Application microservice, user-facing, HTML oriented
graph	Graph Microservice, JSON microservices over an in-memory rdflib graph

These are located in the app_web, and app_graph directories of this repository.

These are Docker-containerized with the caig_ prefix. Therefore, the container names will be caig_web, and caig_graph.

Implementation Summary

• Python3 is used exclusively in the CosmosAIGraph solution
  • See https://www.python.org
• FastAPI is used exclusively as the framework for the web and http services
  • See https://fastapi.tiangolo.com
• Azure Cosmos DB is used as the persistent datastore for source data and session history as well as a vector database
  • One or more Cosmos DB APIs may be part of your solution
  • See https://learn.microsoft.com/en-us/azure/cosmos-db/
  • You can use either the Cosmos DB for MongoDB vCore or Cosmos DB for NoSQL
    • https://learn.microsoft.com/en-us/azure/cosmos-db/mongodb/vcore/
  • Your domain data, with embeddings, are stored here
  • AI sessions, prompt history, feedback, and completion history is persisted here
  • This Cosmos DB data can optionally be mirrored to OneLake in Microsoft Fabric
• Azure OpenAI is used for AI and LLM functionality
  • See https://learn.microsoft.com/en-us/azure/ai-services/openai/
• semantic-kernel is use for AI and LLM orchestration
  • See https://learn.microsoft.com/en-us/semantic-kernel/overview/
• rdflib is used as the high-performance in-memory graph
  • See https://rdflib.readthedocs.io/en/stable/
• SPARQL 1.1 is the graph query language
  • See https://www.w3.org/TR/sparql11-query/
• Web Ontology Language (OWL) is the graph schema/ontology definition language
  • See https://www.w3.org/OWL/

---

Quick Start

Clone this GitHub Repository

Open a PowerShell Terminal, navigate to the desired parent directory and execute the following git clone command. This will copy the contents of the public GitHub repository to your workstation.

If you don't have git installed on your system, please see the Developer Workstation Setup page.

> git clone https://github.com/cjoakim/CosmosAIGraph.git

> cd CosmosAIGraph

> Get-Location

The output value from the Get-Location will be a fully-qualified directory path on your workstation. Please set the CAIG_HOME environment variable to this directory path value.

echo 'setting CAIG_HOME'
[Environment]::SetEnvironmentVariable("CAIG_HOME", "...your value from Get-Location ...", "User")

You will need to restart your Terminal for the above command to take effect.

You'll see in a section below that this CosmosAIGraph reference application uses several environment variables, and they all begin with CAIG_.

Provision Azure Cosmos DB and Azure OpenAI

• See Initial PaaS Provisioning

Developer Workstation Setup

• See Developer Workstation Setup

• See Environment Variables

Load Cosmos DB with Library and Config Documents

• See Cosmos DB Document Design and Modeling
• See Load Azure Cosmos DB for MongoDB vCore
• See Load Azure Cosmos DB for NoSQL

Run the Application on your Workstation

• See Local Execution

• See Explore the FastAPI Framework and Endpoint Documentation

• See Understanding the Code

Azure Container App Deployment

• See Deploying the Azure Container App

Screen Shots of the Current Implementation

• See Screen Shots

---

Next Steps: Customizing this Solution for Your Application

It is recommended that CosmosAIGraph Proof-of-Concept (POC) team has the following skillsets:

• A data analyst who is familiar with your input graph data

• A data engineer who can wrangle/transform the raw data into JSON documents for Cosmos DB

• A Python developer with UI skills

• See Customizing this Solution

• See Code Generation

• See the FAQ Page to clarify your understanding of the CosmosAIGraph solution.