Mayowa Kalejaiye
← Back to portfolio
Case Study · AI Product

GigScout

An AI job feed that scans eight boards, scores every listing against your real skill profile, and hands you a strong match and a ready application together—delivered to your phone.

Next.js TypeScript Prisma AI Scoring
GigScout

Context

Freelance job hunting means checking eight different boards manually, scrolling past hundreds of posts to find the handful worth applying to, then writing a fresh pitch for each one from scratch.

What I built

An AI job feed that scans Remotive, Working Nomads, WeWorkRemotely, Reddit, RemoteOK, Himalayas, and Hacker News on an hourly cycle, scores every listing against a user’s real skill profile, and surfaces only the matches worth their time. The free tier runs rule-based scoring across three boards. The paid tier, five dollars a month, unlocks all seven boards, AI-scored fit using the user’s own API key, and AI-generated job kits—tailored resume bullets, an outreach pitch, and ready-to-send message templates built around the specific posting. A Telegram bot pushes fresh matches directly to a user’s phone, flagging elite roles the moment they surface.

Why this approach

Most job board tooling stops at filtering. The bigger cost isn’t finding a fit, it’s the hour spent writing a pitch after you’ve found one. Building the job kit generation directly into the scoring pipeline collapses that gap, so a strong match and a ready application arrive together instead of as two separate steps a user has to do themselves.

Role & scope

End to end: the multi-source scanning architecture, the scoring engine, the BYOK AI integration, the job kit generation, and the Telegram bot.

Result · three weeks live

29
users signed up
2
paying subscribers — 7% free→paid
$10
MRR — first sale on launch day
669
AI-scored matches delivered
236
scoring pipeline runs
10,855
jobs indexed across 8 boards
$0.26
total AI spend, all-time
+374%
scan growth, 2-week window

Live for three weeks. GigScout signed its first paying subscriber on launch day and now runs a real freemium funnel—29 users, 2 on paid—while indexing over 10,000 jobs across eight boards. The number I care about most: 669 AI-scored matches delivered for $0.26 of total AI spend. The scoring pipeline is built around a bring-your-own-key model and tight prompt budgets, so unit economics stay near zero as usage climbs—the difference between an AI feature that scales and one that quietly bleeds money.

Every path—sign-in, match, generated job kit, Telegram alert—was tested end to end by hand before launch.