Strikebase — Freelance Bid Intelligence

Freelancers waste hours applying for jobs they won't win — not because they're unqualified, but because they're flying blind. Job boards show listings. ChatGPT can only work with what you give it. Neither tells you how many people are already bidding, whether the client has a dispute history, or what the winning rate looks like this week. Strikebase fixes that. We built an AI agent that scans live freelance platforms and returns a Strike Score — 0 to 100 — for every opportunity, backed by data no one else surfaces. Each score includes number-backed reasons, red flags, and a proposal opening line tailored to the listing and the freelancer's own background. Scoring is personalised. Freelancers set their proficiency per skill (Beginner / Competent / Expert), describe their niche in one line, and optionally link their GitHub or portfolio. Bright Data's Web Unlocker reads those pages. The score reflects their specific competitiveness — not a generic market signal. All four Bright Data tools power the pipeline: SERP API discovers fresh listings across five platforms; Web Scraper API extracts bid counts, budgets, and client IDs; Web Unlocker bypasses bot protection to read client profiles and portfolio pages; MCP Server connects all live data directly into Claude's reasoning layer. Cognee powers the memory layer. Every scan builds a per-user knowledge graph. Future scans query it for prior client intelligence and win patterns — the product gets smarter with each use. AI/ML API powers the chat layer. Freelancers ask anything about a specific opportunity with full context pre-loaded: score, bid count, client history, red flags. Real-time streaming answers. Stack: Next.js · FastAPI · Supabase · Anthropic Claude · Cognee · AI/ML API · Bright Data · Vercel · Render.