Browserbot: an attempt at automating digital tasks

An AI-enabled bot that performs tedious, repetitive digital tasks, by referring to a human walkthrough video. Made for the SOTA Hack.

Setting up the environment

Create anaconda environment

conda create -n bbot python=3.10 -y 

conda activate bbot

Install dependencies

pip install -r requirements.txt

Record yourself performing a task as well as speaking through it by running:

 python screen_record.py

Feed frames of the recorded video and transcribed audio to GPT-4 by running:

 python gpt4_feed.py

After checking the task list from GPT-4, use AgentQL to perform the task:

python agentql_executor.py

Name		Name	Last commit message	Last commit date
Latest commit History 9 Commits
config		config
.gitignore		.gitignore
README.md		README.md
agentql_executor.py		agentql_executor.py
browser_bot.py		browser_bot.py
config_reader.py		config_reader.py
find_bot_id.py		find_bot_id.py
gpt4_feed.py		gpt4_feed.py
requirements.txt		requirements.txt
screen_record.py		screen_record.py
telegram_client.py		telegram_client.py