This project provides an AI agent, "Jobber," designed to autonomously search and apply for jobs online by controlling a web browser. It's targeted at job seekers who want to automate the application process, saving time and effort by handling job discovery and submission based on user-provided resumes and preferences.

How It Works

Jobber offers two implementations: a simpler multi-agent conversational approach and a more scalable finite state machine (FSM) approach. The FSM version relies on structured output from OpenAI models, making it robust but limiting the use of cheaper or open-source LLMs. The conversational approach is more flexible with model choice but potentially less scalable. Both agents control a browser instance via remote debugging to interact with job websites.

Quick Start & Requirements

• Install: poetry install
• Prerequisites: Poetry, OpenAI API key, Langsmith key (optional, can be disabled), Chrome browser.
• Setup: Requires starting Chrome in remote debugging mode (--remote-debugging-port=9222) and configuring API keys in a .env file. Resume and preferences are set in user_preferences.txt.
• Run: python -u -m jobber_fsm or python -u -m jobber
• Demo: Loom video
• Docs: sentient framework

Highlighted Details

• Two distinct agent architectures (conversational vs. FSM) for comparison and scalability.
• Leverages browser remote debugging for autonomous web navigation.
• Requires specific LLM outputs for the FSM approach, impacting model choice.
• Includes evaluation scripts for testing agent performance.

Maintenance & Community

• Active development indicated by the availability of two implementations and ongoing improvements to the FSM approach.
• Community chat available via Discord: https://discord.gg/umgnyQU2K8

Licensing & Compatibility

• The README does not explicitly state a license. This requires clarification for commercial use or integration into closed-source projects.

Limitations & Caveats

The FSM implementation's dependency on structured LLM output restricts the use of certain cost-effective or open-source models. The project is presented as part of an "upcoming open-source framework sentient," suggesting potential for ongoing changes and API evolution.