README

README.md

@@ -1,3 +1,20 @@
-## Genai network client
+# Project GenAi client node
 
-This is a client implementation for GenAI network project. `tauri-app` package contains client node with UI and `client-package` contains a separate NPM package that can be used via CLI or as an NPM library.
+Node is the serving component of the [project](https://paipe.io/) backend: an image generation and pipeline builder infrastructure as a service. Node is hosted on the client side and interacts with the [server](https://github.com/paipe-labs/project-genai/) as follows:
+- node recieves an image generation request 
+- node performs the task generation, using client computing resources
+- node sends the generation result to the server
+
+Node components:
+- `SessionManager` class acts as a coordinator: it establishes a session by creating a WebSocket connection to the server and serves the process of data flow between the server and the node
+- `TasksManager` class is responsible for creating and managing image generation tasks
+- `InferenceServer` class declares the interface of interaction with Inference. Currently the following types are available depending on the type of inference:
+    - `AutomaticInferenceServer`
+    - `ComfyUIInferenceServer`
+    - `VoltaMLInferenceServer`
+    - `TestInferenceServer`
+
+![node](doc/node.png)
+
+# Inference
+Inference is a node part responsible for the image generation process. [Сog-comfyui](https://github.com/fofr/cog-comfyui) and [ai-dock comfyui](https://github.com/ai-dock/comfyui/) are integrated into node component as base Docker images - configuration can be found in `build` directory.