AI / ML Layer

Machine Intelligence

Model-driven wallet clustering, entity resolution and on-chain pattern recognition run over your platform's live data.

What the Machine Intelligence page delivers

This is your platform's AI/ML layer. It runs forensic models over your data to uncover hidden links and suspicious behavior invisible to the naked eye — turning raw wallets and transactions into actionable leads.

The 4 capabilities
Address Clustering

Groups separate wallet addresses secretly controlled by the same person or group — using 4 forensic heuristics (co-spend, change-address, behavioral, peel-chain).

Entity Resolution

Finds duplicate actor profiles in your data that are really the same real-world entity — matched via shared aliases, wallets and attributes — so you can merge them.

Pattern Recognition

Scans recent transactions for laundering structures: peel chains, layering, round-tripping and sudden fan-out / fan-in bursts.

Temporal & Behavioral

Profiles each wallet's operational rhythm — active hours, off-hours activity, round-amount propensity and automation cadence — to tell human operators from bots.

How to use it — step by step
  1. 1

    Each panel below is a separate AI model. They all run over your platform's live data — wallets, actors and transactions already imported.

  2. 2

    Press the button at the top-right of a panel (e.g. "Cluster addresses", "Detect patterns") to run that model.

  3. 3

    Wait a moment while it analyzes. Results appear as cards inside the panel, each with a severity, a confidence score and a plain-language explanation.

  4. 4

    Empty result? It just means the model found nothing significant in the current data (e.g. "not enough wallets to cluster"). Import more wallets or transactions and run again.

  5. 5

    Use the results to drive your next move — open an actor profile, escalate to a case or trace the flagged address in the Entity Graph.

Tip: the Sparkles button on each panel re-runs that model anytime — results always reflect your latest data.

Address Clustering / Entity Resolution

Link addresses controlled by the same actor via four heuristics — common-input co-spend, change-address detection, behavioral/deposit patterns and peel-chain fragmentation.

Clustering Heuristics
Co-SpendCommon-input ownership — addresses spending together as one entity (strongest on UTXO chains).
Change-AddrChange outputs back to self that reuse the sender's wallet-software pattern.
BehavioralRepeated shared DEX/router/bridge usage + temporal correlation (EVM-oriented).
Peel-ChainSystematic fund fragmentation across sequential hops.

Not run yet

Run the model to generate intelligence from your current platform data.

Entity Resolution

Detect duplicate actor records likely representing the same real-world entity via shared aliases, wallets and attributes.

Not run yet

Run the model to generate intelligence from your current platform data.

Pattern Recognition

Scan recent transactions for structural anomalies — peel chains, layering, round-tripping and fan-out/in bursts.

Not run yet

Run the model to generate intelligence from your current platform data.

Temporal & Behavioral Analysis

Profile operational rhythm — hourly concentration, off-hours UTC activity, round-amount propensity and automated cadence — to identify single-operator and bot-controlled wallets across clusters.

Not run yet

Run the model to generate intelligence from your current platform data.