Can an AI Agent Apply to Jobs Better Than You?

Can you actually build an autonomous AI agent to handle the entire job application process from deep diving into job descriptions to optimising your resume and cover letter without fabricating a single line of data? In this session, we put cutting edge agentic workflows to the test. We skip the manual copy pasting in Microsoft Word and instead deploy Hermes Agent and Claude Code to analyse a Transformation Lead role, map skill gaps, conduct an interactive interview, and iterate on a highly tailored resume package. But it’s not all smooth sailing. We run face-first into security risks, high API token costs ($2.50+ per session!), and code bugs that require advanced refactoring using Claude Code (Opus 4.8). Later in the video, we pivot to building background infrastructure: creating a custom remote-control agent connected to Telegram and trying to train it on real-world navigation skills using public transport data (with mixed results across Victoria, NSW, and Queensland data pipelines!). If you want to understand the reality of running autonomous agents locally on your machine the capabilities, the costs, and the absolute chaos of debugging them this breakdown is for you. ⏱️ Timestamps & Key Moments 00:00 - Introduction: The Goal & Switching to Hermes Agent 04:03 - Prompt Engineering: Training an Agent Not to Lie 08:57 - Hermes Agent Overview: Obsidian, GitHub, & Telegram Integration 13:42 - The Hidden Costs: Token Burn & API Expenses (Anthropic vs DeepSeek) 24:01 - Security Deep Dive: Should You Run Agents on Your Main Computer? 26:19 - Case Study: Applying for a Transformation Lead Role at Pepperstone 27:51 - The Agent Interview: AI Grilling Me on Jira, Confluence, & Agile 36:49 - Writer/Critic Loops: Overcoming Automated Screening (ATS) Filters 46:19 - Troubleshooting: When the Coding Agent Breaks 52:06 - Refactoring the App Using Claude Code (Opus 4.8) 58:14 - Two-Column PDF Resume Generation: Ditching Microsoft Office 01:04:15 - Live Feedback: Fixing Design Flaws (and a Potential Business Model?) 01:07:00 - Project #2: Building a Victorian Public Transport Navigation Skill 01:12:58 - Remote Execution: Deploying Autonomous Skills to Telegram 01:15:29 - The Data Nightmare: Dealing with Unreliable APIs & Static Timetables 01:24:16 - Open Data Showdown: NSW vs Victoria vs Queensland 01:30:50 - Environment Setup: Local Directory File Management for AI Agents 01:37:25 - The Verdict: Successfully Routing a Gold Coast Multi-Stop Journey 01:44:08 - Hermes vs Claude Code: Autonomy, Persistence, and Future Potential 🔗 Resources Mentioned & Tools Used Hermes Agent: Open-source agentic harness for autonomous, long-lived background tasks. Claude Code: Advanced code-generation and local directory refactoring. Obsidian / Telegram: External platform integrations for background execution. PTV / Transport for NSW Data: Open-source transit feeds utilized in the skill build. What are your thoughts on agentic workflows? Are you ready to hand over your job hunt to a local AI agent, or do the API costs and security risks give you pause? Drop your thoughts, questions, or your own prompt strategies in the comments below! #AIAgents #HermesAgent #ClaudeCode #ResumeOptimization #PromptEngineering #ArtificialIntelligence #JobSearchHack #Anthropic #CodingAgent