UIPATH-Python Social Media Automation Project

UIPATH integration with vision LLM / API

workflow: For each profile will go with the 2 parts

Part 1:
Input Handling & Screenshot Capture
Read Excel with profile links (FB/IG) via UiPath.

Open browser, take full-page screenshot per profile.

Save screenshot and send to Python script.

Python calls local LLaMA model with image as prompt.
LLM output will be to generate a prompt to will be a second input for the next model Same model, maybe it will be Gemini


** Check opening links from Bio Section. how feasible is it?

Part 2: Use Output for Smart Interaction
Use extracted data to write personalized comments & messages. Maybe Claude AI for it


- continue with same iteration with each profile

Scroll, recent post comment, and DM using behavior mimicking human actions (e.g., delay, scroll, random interactions).

- Report in excel file


Part 3

- In tomorrow's run we need to check each profile if it replied or not.
sending messages is (Implemented needs revision) NO NEED TO CREATE
Store message history and queue follow-ups (next-day logic).
--

Profile switching (session management, proxy handling).
-- 50 messages per profile and switching between them.

Detect human replies and auto-reply using LLM + message logic.











UiPath (RPA)

Python (backend logic)

Local LLaMA (LLM with Vision)

Social media automation (FB, IG) DONE

Personalized behavior & message generation

Session management, proxy rotation