updated README

ai/generative-ai-service/oci_langflow/README.md

@@ -7,6 +7,66 @@ This repository contains the code for a prototype of the integration of OCI Gene
 
 Reviewed: 25.06.2025
 
+## Components for Retrieval-Augmented Generation (RAG)
+In this release (jan 2025), we have included the essential components required to build a robust Retrieval-Augmented Generation (RAG) solution:
+
+1. OCI Embeddings Model
+Based on Cohere, this component generates embeddings for your data, enabling effective search and retrieval.
+
+2. OCI Chat Model
+Powered by Meta or Cohere models, this component handles conversational AI tasks for natural and engaging user interactions.
+
+3. OCI Vector Store
+Built on Oracle Database 23AI, this component provides a highly efficient and scalable vector store for managing and querying embeddings.
+
+## Langflow setup
+To setup an environment with Langflow this is the recommended sequence of steps:
+
+1. Create and activate a "conda environment" using Python 3.11
+```
+conda create -n oci_langflow python==3.11
+conda activate oci_langflow
+```
+2. Install the tool [uv](https://docs.astral.sh/uv/getting-started/) in this environment, using pip
+3. Install Langflow using uv. Follow the instructions [here](https://docs.langflow.org/get-started-installation)
+```
+uv pip install langflow
+```
+4. Install oci Python sdk and update langchain-community
+```
+pip install oci -U
+```
+
+## Setup your environment for OCI
+* Clone the github repository
+```
+git clone https://github.com/luigisaetta/oci_langflow.git
+```
+* modify the `set_env.sh` file. Change the env variable `LANGFLOW_COMPONENTS_PATH` to point your local **oci_custom** directory
+* execute the following command to set the environment variables:
+```
+source ./set_env.sh
+``` 
+* start langflow using the command: 
+```
+uv run langflow run
+```
+
+## Notes
+### Setup: do not install Langflow using directly pip, use uv. 
+There are many components and pip really struggle to manage them correctly.
+As a result the installation is really slow. 
+The preferred way to do (as suggested also in the official documentation) is to use
+uv.
+
+### The **Philosophy**
+The main idea is to provide the code as an easy way that can be used from people. 
+Then, if they want, they can add all the details they really need.
+Therefore, for example, wrappers expose only the mandatory parameters that needs to be customized. 
+For the remaining:
+1. I have followed Langchain defaults.
+2. The code is simple and easy to be customized.
+
 ## Security
 Please consult the [security guide](./SECURITY.md) for our responsible security
 vulnerability disclosure process.