1. Company evidence
TenderMeter reads a bounded set of public pages and turns supported services, industries, locations, languages and certifications into normalized company DNA. Every claim retains its source URL.
Deterministic facts, bounded inference
A qualification pipeline designed to say “no” early.
The expensive part of tender work is not finding notices. It is deciding which ones deserve bid-team time without overlooking a disqualifier.
2. Official tender facts
Current notices come from TED. Notice identity, buyer, value, deadline, CPV and geography remain deterministic source facts throughout the pipeline.
3. Hard filtering
CPV, timing, geography, contract value and exclusions reject obvious mismatches before semantic or model inference spends time on them.
4. Qualification
Semantic matching ranks eligible notices. Local Qwen analyzes only the shortlist and must cite stored company/tender evidence for its explanation and risks.
5. Decision-ready delivery
The final report presents the top three current opportunities with fit score, verdict, reasons, risks, next action and official TED source link.