Unified Subject and Label System
A planned Trustline design direction for organizing subjects, identifiers, events, and decision-grade labels.
The Unified Subject and Label System is on the Trustline roadmap. It describes where Trustline's risk memory is heading: a system that can organize identity, wallet, merchant, agent, transaction, evidence, policy, and outcome data without flattening different entities into one generic table.
Agentic finance creates many subjects. A subject can be a user, an institution, an agent, a merchant, a wallet, an x402 server, a payment requirement, a Verifiable Intent presentation, an AP2 mandate, a treasury workflow, or an underwriting event. Some of these subjects are the same in a high-confidence way. Others only share a weak observation. Trustline needs a way to represent both without overstating identity.
Design Goal
The goal is to build a subject graph and a versioned label system.
The subject graph would track subjects and identifiers within source namespaces. A wallet address on XRPL, an EVM address, a domain, a verified agent identifier, a consumer finance user identifier, and an x402 payTo destination are all different forms of evidence. Some can support high-confidence links. Others remain observations until additional evidence appears.
The label system would track facts, scores, policies, and outcomes with lifecycle states. A label may begin as a candidate, become analytics-only, become decision-grade after validation, or be suppressed when it is no longer reliable. This prevents untested signals from silently affecting production decisions.
Example Labels
| Label family | Examples |
|---|---|
| Identity | identity.kyc_level, identity.verified_agent, merchant.domain_verified |
| Wallet and counterparty | wallet.external_risk_score, wallet.sanctions_hit, merchant.mcc |
| Evidence | vi.present, vi.verified, ap2.payment_mandate_present, trace.present |
| Policy | policy.version, policy.review_required, policy.limit_profile |
| Decision | decision.approved, decision.reviewed, decision.denied, decision.risk_score |
| Outcome | outcome.paid_on_time, outcome.disputed, outcome.settlement_failed |
Only decision-grade labels influence production decisions. Candidate and analytics-only labels support research, monitoring, or product analysis without silently affecting approvals.
Why It Matters
Without a unified subject and label system, every product builds its own partial memory. X402 Secure may know a payment endpoint. Claw Credit may know an agent credit line. Consumer finance workflows may know a user and a merchant. Institutional workflows may know a desk, account, policy, or treasury operation. External signals may know a wallet or entity. These views need to be connected carefully so Trustline can understand repeated behavior across products while respecting boundaries and confidence levels.
The system is also relevant to liability. A later review should be able to understand which subject acted, which policy governed the action, which labels were decision-grade, and which observations were still provisional. That structure helps Trustline support serious financial workflows without pretending that every identifier is equally reliable.
The system is also designed to avoid risky shortcuts. Weak identifiers remain observations until stronger evidence supports a link. Labels move into decision-grade usage only after validation. Deletion and suppression events propagate to affected labels. Rollout begins with shadow writes and analytics before production decisions depend on new labels.
Rollout Principles
The design direction follows conservative rollout principles.
- New tables and adapters can be introduced without changing existing production decisions.
- Source adapters write observations only after the source event commits successfully.
- Decision enrichment remains disabled until shadow data quality is proven.
- Rollback remains configuration-driven, without requiring a risky data migration.
- Deletion and suppression semantics are part of the design from the beginning.
This roadmap is included because it is central to Trustline's future. A mature underwriting system needs durable memory. The Unified Subject and Label System is the planned structure for that memory.