Radar

ob searching is broken at a structural level. Candidates spend 40+ hours per week on a process that is entirely manual, fragmented, and blind. They scroll through 10+ platforms separately, rewrite CVs from scratch for every role, submit applications into black holes with zero feedback, walk into interviews unprepared, and never learn why they keep getting ghosted. There is no tool on the market that connects these stages into a learning system. Existing solutions (LinkedIn, Indeed, Jobscan, Teal, Kickresume) each address one narrow slice — none close the loop between “what I did” and “what worked.”

How We Transformed Experimental AI into a Venture-Ready Product
Radar started as a set of experimental Python scrapers and a basic Claude API wrapper. Over this 4 sprints, we transformed it into a production-grade autonomous job search platform with four distinct agentic AI systems, a multi-tenant backend.

Sprint Day 1 — Built the scraping infrastructure: 9 French job platform connectors (France Travail API, APEC, Welcome to the Jungle, Indeed, LinkedIn, JobTeaser, Malt) running through a Node.js Express proxy with cron-based automation, job matching pipelines, and a pending approvals workflow. Established the Supabase PostgreSQL backend with Row Level Security for multi-tenant data isolation.

Sprint Day 2 — Shipped the core AI layer: Claude-powered job scoring and matching, profile analysis, and the LaTeX CV generation pipeline with HTML preview. Built the onboarding flow, authentication system, theme system, and Chrome extension architecture for passive application tracking.

Sprint Day 3 — Delivered the CV Package Builder: a 4-step agentic workflow from raw profile to complete application package. Built the CV Intelligence Agent (conversational achievement extraction), ATS Gate (keyword gap analysis with predicted callback rates), AI cover letter generation, and Hiring Manager Discovery via LinkedIn X-Ray search with AI-drafted outreach messages.

Sprint Day 4 — Shipped the two features no competitor has. The Agentic Interview Coach: a 5-agent pipeline where candidates select an application they’ve made, and autonomous agents research the role, plan adaptive questions, conduct a mock interview with voice I/O, independently evaluate performance across six dimensions, and generate a personalised coaching plan. The Application Intelligence Agent: a 4-agent analytics pipeline that learns from your entire application history — tracking response rates by platform, CV version, score range, and contract type — then generates strategic insights and outcome predictions for pending applications.

Market & Traction
Target market: 15M+ active job seekers in France alone; 200M+ globally
Monetisation: Freemium (limited scrapes + basic CV) → Pro (unlimited platforms, all agents, intelligence analytics, HM discovery) via Stripe integration (already wired)

Competitive gap: We analysed 50+ competing tools (LinkedIn, Indeed, Jobscan, Teal, Kickresume, Huntr, Simplify, LazyApply). Zero offer the outcome feedback loop. Zero connect interview prep to actual application data. Zero discover hiring managers as part of the CV workflow

Sprint traction: Full working product with 9 platform integrations, 4 autonomous agent systems, and a demo environment with realistic data covering 12 jobs, 8 applications across all status stages, and a complete candidate profile