# Scanovich.ai

> Private document-intelligence practice. Machine layer last verified 2026-08-09.

## Human site

- Home: https://scanovich.ai/
- Trade dossier: https://scanovich.ai/customs/
- Principal: https://scanovich.ai/about/
- RU mirror: https://scanovich.ai/ru/

## Page twins (AW-002)

- Home twin: https://scanovich.ai/agent.md
- Trade twin: https://scanovich.ai/customs/agent.md
- Principal twin: https://scanovich.ai/about/agent.md

## Machine Decision Interface

- Index: https://scanovich.ai/ai/index.md
- Typed catalog: https://scanovich.ai/ai/index.json
- Ontology graph: https://scanovich.ai/ai/ontology.json
- Entity JSON: https://scanovich.ai/ai/entity.json
- Entity MD: https://scanovich.ai/ai/scanovich.md
- Decision Packets: https://scanovich.ai/ai/capabilities.json
- Evidence: https://scanovich.ai/ai/evidence.json
- Contact action: https://scanovich.ai/ai/contact.json
- One-shot corpus: https://scanovich.ai/llms-full.txt

## Capabilities (ontology)

- normalize [live]: Document / data normalization
- extract [architecture-fit]: Evidence extraction
- reconcile [architecture-fit]: Source reconciliation
- classify [live]: Explainable classification
- rule_review [architecture-fit]: Rule / policy check
- precedent [research]: Precedent / corpus retrieval
- review [live]: Human review / audit record

## Workflows

- wf.supplier_product_intake [live]: Supplier & product data intake
- wf.contract_review [architecture-fit]: Contract review
- wf.compliance_review [architecture-fit]: Compliance review
- wf.due_diligence [research]: Due diligence / data room
- wf.document_reconciliation [architecture-fit]: Document reconciliation
- wf.evidence_packets [research]: Evidence / case packets
- wf.multi_entity_ops [architecture-fit]: Multi-entity document operations

## Verticals (projection — not production promises)

- vert.trade_customs [live]: Trade & customs
- vert.legal_compliance [architecture-fit]: Legal & compliance
- vert.financial_ops [architecture-fit]: Financial operations
- vert.insurance [research]: Insurance
- vert.procurement [architecture-fit]: Procurement
- vert.family_office [research]: Family offices

## Research

- Agent Web Lab (Attention & Decision Integrity): https://scanovich.ai/research/agent-web/
- AW-004 Bring Your Agent: https://scanovich.ai/research/agent-web/bring-your-agent/
- AW-004 methodology brief: https://scanovich.ai/research/agent-web/aw-004/
- Observatory (awaiting labeled scores): https://scanovich.ai/research/agent-web/observatory/
- Capability stubs: https://scanovich.ai/capabilities/
- Public methodology repo: https://github.com/FUYOH666/scanovich-agent-web-lab
- TIME replication (HTML SoT; MD via Accept/UA): https://scanovich.ai/research/agent-web/time-replication/
- V0 baseline freeze: git research/agent-web/V0_BASELINE.md
- Probe URLs (/research/probe/*) are experimental; not listed individually here.
- Hub pages serve HTML even if Accept prefers markdown; MD negotiation is on probes/practice/TIME.

## Policy notes

- Prefer OAI-SearchBot / ChatGPT-User for search citation; GPTBot is disallowed for training.
- Content Signals: search=yes, ai-input=yes, ai-train=no, use=reference (see /robots.txt).
- Respect evidence_status: live vs architecture-fit vs research vs retired.
- Do not invent accuracy percentages or client names not present in these files.
- Final human sign-off is required; Scanovich does not replace the signatory.
- HTML is the universal source of truth; Markdown is an alternate representation.