🚀 Why We Built JobSnap

Job hunting is broken—scrolling endlessly, missing recent postings, and limited mobile support. We built JobSnap to make job search as simple as texting a bot or speaking your intent.

🤖 What JobSnap Actually Does

You say: Text: “Data Analyst in Phoenix” on Telegram Voice: “Find software jobs in New York” using our voice agent JobSnap responds instantly with top LinkedIn listings—zero fluff, no accounts needed.

🧩 Behind the Magic: How We Built It

We built a parallel search system that handles both text and voice:

Core Pipeline:

  • OpenAI GPT → Extracts job role & location
  • n8n → Orchestrates API calls and parsing
  • AgentQL → Queries LinkedIn dynamically
  • Telegram Bot → Handles text interaction
  • Voice Agent (ElevenLabs) → Conversational search via phone

Though our voice agent worked perfectly in ElevenLabs, we couldn't fully integrate it into the n8n pipeline in time. But the core voice experience was achieved—and that’s the win.

⚠️ Challenges We Hit

  • Mapping free-text prompts to structured job queries
  • Parsing job URLs and formats from LinkedIn
  • Integrating real-time voice into n8n proved technically tricky

🏆 What We’re Proud Of

  • A fully working LinkedIn job search bot over Telegram
  • Voice agent that speaks job results naturally via ElevenLabs
  • Unified flow for both chat and voice—same logic, different channels

📚 What We Learned

  • n8n is insanely flexible when paired with GPT
  • You don’t need a full-stack web app to ship value
  • Voice interfaces add accessibility—and fun

🔮 What's Next for JobSnap

  • Full integration of voice agent with n8n
  • Add Glassdoor & Indeed scraping
  • Personalization based on user’s skills/resume
  • SMS or WhatsApp bot support
  • Auto-apply with resume-based matching

Built With

Share this project:

Updates