Solo practice · cross-border commerce AI · pilots open

Explainable AI copilots
for cross-border commerce.

Scanovich scopes explainable AI to operational constraints in cross-border commerce — supplier spreadsheets, marketplace exports, customs archives, settlement attestation. Each system is built to what must hold before someone signs off.

Working products, not slides

Four production surfaces.

Each one is narrow enough to describe in one paragraph — customs, marketplaces, and settlement infrastructure for RWA.

Live

TN VED / HS classification bot

Telegram bot for importers and brokers. Given a product description, it returns an explainable HS code: classification reasoning (GRI 1–6), confidence, and source references. EAEU TN VED today; architecture generalizes to any HS-based jurisdiction.

Try the bot on Telegram
Live

Supplier spreadsheet normalizer

Supplier files arrive in Excel, CSV, PDF, sometimes as photos. The normalizer maps rows to a single schema, translates foreign-language descriptions, prepares fields for HS classification, and flags mandatory vs. voluntary certification. Built for importer ops teams who lose hours on cleanup before classification can even start.

See the customs page
Live

YM Analyzer — P&L copilot for Yandex.Market

Live SaaS at app.scanovich.ai. Sellers upload raw exports from their marketplace cabinet, and get SKU-level profit after fees, logistics, storage, and ad boosts. Excel output, period comparison, no manual mapping.

Open the product
SEABW 2026

AttestRWA — settlement attestation for RWA

On-chain compliance bridge for stablecoin real-world-asset settlements. Bank-grade release via EAS attestations and programmable escrow — not property tokenization. Presented at SEA Blockchain Week 2026; live on Base Sepolia testnet.

View project

Why one company, not two

Both wedges turn operational mess into explainable AI.

Importers and marketplace sellers live with the same problem in different clothes: raw, inconsistent, hard-to-trust operational data that blocks decisions. Scanovich builds the same thing for both — AI workflows that normalize the data, explain their output, and stay auditable for the person who signs off.

Customs AI stack

For importers and customs brokers

Dealing with supplier data, HS classification, and declaration prep — at a speed that stays compatible with what a reviewer can defend.

  • Supplier spreadsheets → clean, reviewable schema
  • Explainable HS / TN VED classification with GRI trace
  • Certification requirement detection: mandatory / voluntary / none
  • Archive search over released declarations (in pilot)
Customs product page
Marketplace copilots

For sellers who need real economics

Not cabinet dashboards. SKU-level P&L after fees, logistics, storage, and ad boosts — with a path to Amazon and Shopify once the fee models are added.

  • SKU-level profit, not aggregate margin
  • Excel-ready output, period comparison
  • YM Analyzer live; Ozon and Wildberries in design-partner search
  • Same engine, portable to any marketplace with a fee model
Marketplaces hub

How it works

Paste a product description → receive a classification you can defend.

Every output ships with the reasoning trace, confidence, and the regulatory sources used to get there. The person signing off sees what the system saw — not just the final code.

The same loop runs on spreadsheet rows at scale: translate, normalize, classify, flag certification requirements, and hand back a file that's ready for the next step.

Illustrative output — current bot response format
Input
“LED desk lamp, 10W, USB-C, touch dimmer”
HS / TN VED
9405 42 390 0
Confidence
0.86
GRI trace
  • Rule 1 → Heading 9405 (lamps and lighting fittings)
  • Rule 6 → Subheading 9405.42 (LED, electric)
Certification
EAEU TR 004/2011: mandatory declaration of conformity
Sources
  • · EAEU TN VED, Chapter 94
  • · TR CU 004/2011, Annex 1

Practice

Systems shaped to the operating problem — not the slide deck.

Scanovich is a solo practice in applied AI for cross-border commerce: customs workflows, marketplace unit economics, settlement attestation. Each engagement starts from your decision surface — what must hold before someone signs off — and ends in software that is explainable, testable, and scoped to that constraint.

Engagements typically take one of three forms: a bounded pilot on your data under NDA, a design-partner seat while a new wedge is defined, or advisory input from operators who already know the problem from the inside. Investor conversations follow once pilot signal exists; the first step is a short, concrete email.

Contact

Talk to us

Questions about pilots, partnerships, or deployment? We usually reply within a business day.

Direct

JST (UTC+9). Telegram — same day. Email — within 24h.