Procurement Technology / AI
TenderScanner
AI procurement intelligence platform
My own SaaS product: scrapers ingest public tenders across Serbia and Bosnia and Herzegovina, and an AI pipeline analyzes each one against company profiles to produce go/no-go recommendations.
Contribution
- Everything - product, frontend, backend, data platform and AI pipeline, built solo
- Scrapers ingesting newly published tenders from public procurement sources
- AI analysis matching tenders to company profiles for go/no-go decisions
Companies bidding on public tenders in Serbia and Bosnia and Herzegovina track them the hard way: fragmented procurement portals, manual checking, and a real cost to missing the one tender that fits. TenderScanner is my answer to that - a product I design, build and run solo.
The pipeline works end to end in production: scrapers ingest newly published tenders, and a Gemini Flash–based analysis step processes each one so the system can match it against a company’s profile and produce a go/no-go recommendation with supporting detail. A React and TypeScript frontend sits on a .NET backend with PostgreSQL underneath.
The product is live and working, with the first interested clients and a feature roadmap ahead - and it doubles as the project where I own every decision, from scraper resilience to what an AI recommendation owes its reader.
Outcomes
- Deployed and operational - scrapers and AI analysis run on every newly published tender
- First client conversations underway while development continues