What "transparent AI matching" actually means
We show why the AI picked (or didn't pick) you for a project. Here's the shape of that reasoning — and why opacity is dangerous.
Every major hiring platform uses AI to rank candidates. Almost none of them explain the ranking. This is a bug, not a feature.
Why opacity is dangerous
- Candidates can't improve without feedback.
- Hiring managers can't audit for bias.
- Platforms can slip commercial preferences into "AI recommendations" and no one notices.
We built ScopeCred's AI Match with the opposite bias: show your work.
What you actually see
Every match returns a structured payload:
- Score (0-100). Anchored to Web3 project fit specifically.
- Strengths. Cited from your profile: verified skills, past project types, trust history.
- Gaps. What's missing or weak. Not a rejection — a checklist.
- Verdict. One sentence summarising the trade-off.
Both sides see it. Contributor sees "you're an 85 for this project because…". Owner sees "this contributor is an 85 because…". Same reasoning, symmetric transparency.
How we keep it honest
- Model is public: GPT-4o-mini via Emergent LLM Key.
- System prompt is public (in our repo).
- Cache TTL is 5 minutes — refresh on profile change.
- Score is deterministic given the same input, so identical inputs produce identical outputs.
The point of all this
An AI-transparency system that shows both sides the reasoning is a negotiation aid, not a gatekeeper. Contributors bid smarter. Owners refine briefs. Trust compounds.
That's what "transparent AI matching" means, and why it should be table stakes.