Back to Case Studies
Case Study

Scout Jobs

AI-Powered Executive Job Search Agent

Automated job scraping, intelligent filtering, Chrome extension, multi-format resume generation, and a full-stack dashboard, built by a non-developer in 10 days.

5,583
Lines of Code
17
API Endpoints
10
Days to Build
$200
vs $222K

The Comparison

The same scope estimated through a SAFe 6 Agile Release Train: 9 people, 1 Program Increment, 5 sprints of execution plus pre-PI ramp-up.

Metric
Solo + Claude CodeWhat happened
SAFe 6 Enterprise9-person ART estimate
Team Size
1 person + AI9 people (6.75 FTE)
Elapsed Time
~10 days12-14 weeks
Total Labor Hours
~35 hrs~2,700 hrs
Hours in Meetings
0529 hrs (20%)
Documents Produced
0~160
Deployments
Continuous (same day)1-2 per sprint (gated)
Decision Latency
SecondsDays to weeks
Cost
~$200~$222,000

10-Day Build Timeline

From zero to production while actively using it for a real job search

Timeline reconstructed from file creation/modification dates.

Days 1-2March 2-3

Core application: Python HTTP server, SQLite database, Google OAuth authentication, job CRUD endpoints, and the initial web dashboard with search and filtering.

Days 3-4March 4-5

Indeed job scraping automation with 3 targeted search queries running 3x daily. Smart knockout filter system with configurable rules for title, location (50-mile radius), and salary ($250K+ threshold).

Days 5-6March 6-7

Chrome extension (Manifest V3) for one-click job clipping. Claude AI integration for intelligent extraction of structured job data from raw HTML.

Days 7-8March 8-9

Multi-format resume generator with ReportLab PDF output: 5 visual formats and 3 role-specific content versions. In-app resume version editor for customizing sections, summary, and skills.

Days 9-10March 10-12

LinkedIn integration via AgentMail API for monitoring job alert emails. Rejection tracking, deduplication, GCS database sync, Docker containerization, and Cloud Run deployment.

What Was Built

Automated job discovery: Indeed scraping 3x daily across 3 queries + LinkedIn email monitoring via AgentMail

Smart filtering engine: Configurable knockout rules for location (50-mile radius), salary ($250K+), and title exclusions

Job pipeline dashboard: Full web UI with search, sort, pagination, status tracking, notes, and application status

Chrome extension: One-click job clipping that uses Claude AI to extract structured data from any job posting page

AI-powered job parsing: Anthropic API extracts title, company, salary, location, and requirements from unstructured HTML

Multi-format resume generator: 5 visual formats and 3 role-specific versions generated from a master JSON resume

Cloud-synced database: SQLite auto-syncs to/from Google Cloud Storage on every server start and mutation

Containerized deployment: One-command deploy to Google Cloud Run via Docker

Google authentication: Google Sign-In with email allowlisting and signed session cookies

Rejection tracking: Automated handler updates job status and syncs with Teal ATS on rejection emails

Tech Stack

LayerTechnology
BackendPython 3.12 (stdlib HTTP server, zero framework)
FrontendVanilla JavaScript + HTML (embedded SPA)
DatabaseSQLite (with GCS sync)
AIAnthropic Claude API (job parsing, extraction)
AuthGoogle Sign-In (OAuth 2.0) + signed cookies
ResumeReportLab (PDF), Node.js CLI templates
ExtensionChrome Manifest V3, vanilla JS
Job SourcesIndeed (browser automation), LinkedIn (AgentMail API)
InfrastructureDocker, Google Cloud Run, Google Cloud Storage
DeploymentShell script → gcloud builds submit → Cloud Run

Multi-Component Architecture

Web DashboardFull CRUD, search, filtering, pagination
Scraping EngineIndeed 3x daily + LinkedIn via AgentMail
Chrome ExtensionManifest V3, popup + content scripts
Resume Generator5 formats x 3 versions, PDF output
Knockout FilterConfigurable rules engine
“Scout Jobs proves that one non-technical executive with Claude Code can ship a production-grade, multi-component application in 10 days and ~15 hours of personal effort, delivering the same output that a 9-person SAFe team would spend 14 weeks, 529 hours of ceremonies, 160 documents, and $222,000 to produce, with the majority of that enterprise investment going not to building software but to the process of talking about building software.”
Rick Becker