ENGINEERING / ARCHITECTURE
TECHNOLOGY

Built on evidence.
Designed for understanding.

Wallet intelligence should start with verifiable blockchain facts, not confident-sounding guesses. Here is how we are building toward that standard.

THE PIPELINE

From on-chain evidence to plain English.

A deliberately simple architecture with clear boundaries between data, interpretation, and product experience.

01 / 04

Public wallet address

A user supplies a public Ethereum address. No private keys, wallet connection, or signatures are requested.

02 / 04

Blockchain data retrieval

In live mode, the backend queries Ethereum transaction and receipt data through a provider.

03 / 04

Evidence normalization

Deterministic fields are organized before being passed to an AI model. On-chain metadata is treated as untrusted input.

04 / 04

Human-readable explanation

Claude can explain verified evidence and identify uncertainty. Demo mode uses prewritten examples instead.

SYSTEM PRINCIPLES

Clarity is a technical requirement.

Read-only by design

We never ask for a seed phrase, private key or permission to sign a transaction.

Traceable evidence

Transaction hashes and source records are the foundation of live explanations.

Explicit limitations

Interpretation is not proof of intent, identity, security or financial outcome.

CURRENT IMPLEMENTATION

One product. Two execution modes.

We can demonstrate the interaction model today without implying that simulated output is live AI analysis.

Try the demo
AVAILABLE
Demo mode

Fictional wallet fixtures and prewritten transaction explanations. No external API calls.

IN DEVELOPMENT
Live mode

Ethereum provider and Claude integrations exist as a prototype; production safeguards and evaluation remain outstanding.

Building the intelligence layer for Web3.

See what we are prioritizing next.

Explore roadmap