Building and Monetizing a Job Search Tool with Claude

EPD11183 · reddit · 2026-07-17

After losing his job, the author used Claude to build a job search tool from scratch, eventually turning it into a real product.

Key Process

First, he had Claude organize his resume, generate a master resume, and act as a career advisor to flesh out his experience and suggest job directions.
Next, he had it write a web scraper to extract job postings from fintech ATS pages into a CSV file.
He then iterated on a job scoring system: filtering by title, description, and negative keywords, later turning it into a GUI with adjustable weights.
He also added an application management CRM, Kanban-style tracking, interview prep, and Gmail sync.

Architecture Evolution

Initially built with Streamlit + SQLite for rapid prototyping.
Later, to support multiple users and productize, the backend kept Python/FastAPI, the frontend moved to React/TypeScript, and the database migrated to Postgres.
Deployed on Cloudflare, Railway, and Hostinger.

About Fable

He noted that after Fable was released, his workflow evolved further: he first had it do research and planning, then spin up multiple subagents to implement features and review code, before deploying it himself. The author considers this a critical leap from "AI helping write scripts" to "AI planning, dividing labor, and reviewing."

Results

The tool also features an analytics dashboard to track:

Which score ranges yield higher response rates
Which resume version performs best
Whether referrals are effective
If applying within 48 hours provides an edge
Which keywords correlate with interviews/screening

The author ultimately found a new job in 9 weeks and transformed the product into a "search result scoring + proactive push" job search tool; it currently offers a free trial and optional paid tiers.

Original post →

Building and Monetizing a Job Search Tool with Claude

More from coding & agent