Chain adapters
Normalize BTC, ETH, ERC-20 USDT/USDC, and BLCH events while preserving each network’s asset and finality rules.
BLOCH INC / DATA INTELLIGENCE
A MATTER OF DATA 2.0
See the signal.
Keep the context.
Graphus is the proposed evidence graph behind Bloch Inc intelligence services. It connects permitted, finalized chain observations with source history, confidence, and human review across BTC, ETH, Ethereum ERC-20 USDT and USDC, and BLCH.
Product preview · Synthetic wallets · No live wallet scoring
GRAPH VIEW / 001 SYNTHETIC
TRACEABLE RELATIONSHIPS03 NETWORKS / 05 ASSETS
01 / PRINCIPLE
Context changes
what data means.
A transaction, a source record, and a graph path are different kinds of evidence. Graphus keeps them distinct, dated, and inspectable so analysts can see a reason for review without treating proximity as guilt.
02 / THE SYSTEM
From raw events
to accountable insight.
Six linked layers make intelligence reproducible and reviewable. Each is designed to operate outside chain consensus and behind private access controls.
Source lineage
Carry source, license, timestamp, confidence, and expiry with every external claim. A stale signal stops at the gate.
Temporal graph
Trace bounded value-flow paths, verified bridge receipts, and changing relationships. Shared infrastructure remains a confounder.
Measured models
Evaluate only reviewed outcomes with time-separated tests, error costs, calibration, and rollback before any promotion.
ZK research
Test proofs for small deterministic feature transforms or inference. Proof validity cannot establish source truth or model accuracy.
Human decisions
Show evidence paths and alternatives, record analyst decisions, and allow correction before consequential customer action.
03 / INITIAL ASSET MAP
One graph.
Five asset lenses.
Explore the proposed event model for the first five assets on three networks. The map describes architecture; current Ethereum and Bitcoin jobs check source heights only, while Bloch has a bounded indexer sample.
UTXO / PROBABILISTIC FINALITY
DESIGN / NO TRANSACTION INGESTIONBitcoin flow map
01Block + transactionSOURCE EVENT
→02Outpoint + spendCANONICAL FACT
→03UTXO flow graphTYPED RELATION
→04Temporal motifsREVIEW SIGNAL
Inputs spend earlier outputs; multiple inputs do not establish a single owner. Change and CoinJoin heuristics remain reversible hypotheses.
Boundary: Bitcoin tip height is observed; confirmation policy and full UTXO reconstruction are future gates.
Illustrative system map. A graph edge is not an identity claim, a sanctions decision, or proof of illicit activity.
04 / GRAPH INTELLIGENCE
Every edge needs
a reason to exist.
Select an example to see how Graphus would distinguish observations from claims. These examples are synthetic design illustrations.
OBSERVATION→PROVENANCE→REVIEW
FINALIZED EVENT · SOURCE ID · EXPIRY
Trace a path, retain the caveat.
A bounded, directed path can help prioritize review. The path is an association with source references and time limits; it is never a verdict about a person or wallet.
Association only · Human review required
05 / ANALYTICS WORKBENCH
Depth with
an audit trail.
The research roadmap combines deterministic graph methods with measured ML. Every output needs its inputs, source time, uncertainty, and an alternative explanation.
Flow and motif search
Time-ordered paths, pass-through, fan-in and fan-out, peel-chain candidates, cycles, bridge round trips, and shared-service hub warnings.
Prototype: bounded paths with finality, expiry, and bridge-proof gates.Change and anomaly
Per-chain and per-asset baselines, burstiness, concentration, dwell time, peer comparison, and missing-data-aware anomaly signals.
Roadmap: review leads, never illicit labels.Temporal graph ML
Compare time-aware embeddings and graph models with transparent rule baselines. Use reviewed labels, entity-disjoint holdouts, calibration, and drift checks.
Current: validation gates only; no trained AML model.Evidence replay
Reconstruct the exact path, source snapshot, block version, feature transform, and reason code behind an analyst view or model output.
Roadmap: reorg replay, correction, and dispute propagation.Purpose-bound access
Separate identity and IP context from chain observations; use tenant-scoped pseudonyms, short retention, field rights, and private audit logs.
Current: prototype privacy helpers; legal review required.Bounded ZK proofs
Benchmark verifiable feature computation or small inference against committed inputs and pinned code, with proof cost and leakage measured.
Research: no prover or proof deployed.06 / APPLICATIONS
One evidence fabric.
Purpose-specific services.
Graphus is a shared technical layer. Each customer use has its own lawful purpose, rights, access boundary, and retention rule.
FIRST SPECIALIZED SERVICE
BI-PoRB
Bloch Inc Proof of Red and Blue Score uses Graphus for transaction signals, wallet context, official-list evidence, and reviewer workflows. Results are advisory, tenant scoped, and external to Bloch Protocol.
View BI-PoRB previewFUTURE / GOVERNED
Market data context
Documented flows and asset activity for authorized market infrastructure research, with aggregation and source rights matched to the use.
FUTURE / GOVERNED
Network observability
Track chain adapters, source freshness, reorgs, and data quality without exposing wallet-level findings on public pages.
05 / TRUST BOUNDARY
Private by design.
Evidence by default.
No on-chain labels. Graphus does not change validation, transaction acceptance, blocks, or balances.
Minimized personal data. Identity and IP context remain separate, restricted, and purpose limited.
Reviewable signals. Disputes, expiry, and source updates can downgrade or remove a finding.
No live claims yet. Current source jobs check health and small samples; no production model or ZK proof is deployed.
06 / ROADMAP
Build the evidence.
Then earn the claim.
- NOW
Private foundation
Dedicated data and ML hosts, OFAC sync, Bloch sample probe, and source-height checks for Ethereum and Bitcoin.
- NEXT
Initial asset adapters
Rights-reviewed ingestion, bounded storage, replay, reorg handling, and finality verification.
- VALIDATE
Graph and model quality
Reviewed labels, temporal holdouts, false-positive costs, calibration, bias tests, and analyst appeal flow.
- RESEARCH
Verifiable computation
Benchmark a small ZK proof for feature transforms or inference before deciding whether it belongs in a service.
GRAPHUS OF BLOCH INC
A Matter of Data 2.0
Better intelligence begins with a better account of what the data can actually support.