LeadScout - AI-Powered Local Lead Generation
From a freelancer's weekend prospecting pain to a public SaaS: AI-driven discovery, crawling and outreach for the DACH region
The Challenge
Like many freelancers and small consultancies, the founder spent entire weekends on manual prospecting: Googling local businesses, copying imprints into spreadsheets, eyeballing whether a website looked outdated. Hours of work for a handful of half-decent leads, with no reliable signal of who actually had a digital need - and every blind cold email chipping away at the sender's reputation.
The Solution
LeadScout was built first as a personal automation, then opened up as a SaaS product when friends asked to use it. A FastAPI backend orchestrates a Celery pipeline that discovers companies via Google Places (PLZ + radius search powered by PostGIS), crawls their websites with Playwright, detects the technology stack, audits SEO, checks social media presence, and runs a structured AI analysis to score digital maturity, identify weaknesses and draft personalized outreach. A Next.js dashboard lets users browse, sort and act on prioritized leads. Hosted in the EU on Supabase, GDPR-compliant from day one.
Our Approach
Personal Automation Experiment
Started in 2025 as a self-serve tool: AI for analysis, Google Places for discovery, Python glue and Playwright for crawling. Within weeks the tool produced better leads than a full month of manual research.
Two-Phase Pipeline Architecture
Split discovery (cheap Google Places call) from analysis (expensive crawl + GPT-4o) into separate Celery tasks, so users see results immediately and only spend AI budget on companies they actually care about.
Geo Search with PostGIS
Built a German postal-code geodata layer with PostGIS ST_DWithin radius queries - users enter a PLZ + radius (5-50 km) and the system finds every relevant business within range.
Structured AI Analysis
GPT-4o prompts use response_format=json_object with an explicit verification loop that grounds claims in the SEO audit and tech detection results - preventing the model from inventing weaknesses that don't match the data.
From Solo Tool to Public SaaS
Cleaned up the codebase, added Supabase auth (EU region Frankfurt) with asymmetric JWKS verification, wired Stripe billing, built a marketing site with i18n (DE/EN), and shipped GDPR-compliant by default.
Transparent Pricing & Direct Support
Independently funded, no growth-chasing investor pressure. Pricing covers actual API costs (~0.03 EUR per lead in OpenRouter + Google Places) plus hosting and maintenance time. Every email lands directly with the founder.
The Results
From days of manual Google research to a sorted, scored list of prioritized leads in under 30 minutes.
AI-grounded weakness signals replace gut-feel judgments - users reach out with concrete, verifiable observations.
Postal-code search across Germany (8,000+ PLZ codes), with Austria and Switzerland on the roadmap.
Conclusion
LeadScout shows what happens when a real practitioner pain (weekend prospecting) meets AI-augmented engineering: an automation experiment becomes a sustainable, independently funded SaaS that respects user data and gives freelancers and small agencies in the DACH region the lead intelligence that previously required a sales team.