Work

Evidence-backed AI

Vettique

I connected seven evidence sources to Vettique's AI supplier-risk reports and separated missing information from signs of risk.

Vettique supplier-risk report with an evidence checklist, unavailable sources, assessment notes and shipment records.
396,339misattributed shipments removedCaught and removed before launch.
Client
Supplier due diligence
Role
Integrated
Timeline
2026 · Delivered
Stack
Next.js 16, React 19, Supabase, Gemini 2.5, ScraperAPI, Google Maps & Search, ImportYeti, WHOIS, Google Safe Browsing

Patrick is an excellent developer and a genuinely trustworthy partner. He delivered a complex data integration milestone involving 7 external sources, built a clean pipeline architecture, and was honest about technical limitations rather than overpromising. He communicates proactively, fixes issues swiftly, and I'm continuing to work with him on a second project. Highly recommended.

Ziad Ayad, Founder, VettiqueSupplier due diligence

Seven external evidence sources feed the AI-written report. Missing data has a separate status, and verified manufacturers receive evidence-based scores.

A language-model-only score could label a legitimate manufacturer as unsafe because evidence was missing.

Missing information inflated supplier risk, and misattributed shipment data could contaminate reports.

  1. Connected seven sources of supplier information and rewrote the AI’s instructions so the evidence drives its risk analysis.
  2. Changed the reports to distinguish missing information from evidence of risk and show which checks completed.
  3. Removed shipment records that had been assigned to the wrong suppliers.

Separate unknown from unsafe

Keep data gaps in their own labeled section so missing evidence cannot manufacture risk.

Expose coverage

Show which checks ran, were blocked, or found nothing, rather than presenting every result as equally complete.

What the project delivered

7sources of supplier informationCustoms records, supplier profile, category pricing, business listing, domain age, scam and safe-browsing checks, web presence.
85 to 25One supplier’s risk score changed from 85 to 25After correcting the evidence used in that assessment. This is one recorded example.

Say hello.

Tell me what you'd like to build or improve.

Quick navigation

Find a page.

Use the arrow keys to choose a result, Enter to open it, or Escape to close.