Beta — fifty design partner slots

The conversation
is the attack.

A behavioral analysis layer for your crypto messages. Cyphervex reads every Telegram, Discord, and X DM in real time and filters the scams — pig butchering, fake support, impersonation, deepfake setups — before they reach you.

Request access→Try the live read
001 — The thesis

AI moved the attack upstream.

The losses happen in Telegram DMs, Discord pings, X replies, and Calendly links — long before any transaction is signed. Blockaid sees the transaction. TRM sees the laundering. Chainalysis sees the cluster. We read the message that started it all.

$17B
Global crypto scam losses in 2025
Chainalysis Crypto Crime Report 2026
4.5×
More profitable when scams use AI
Chainalysis · 2026
1,400%
Year-over-year growth in impersonation scams
Chainalysis · 2026
002 — The read

Every message read against a behavioral baseline. In real time.

cyphervex / inbound / telegram
Live
> message_received {
from: @alex_paradigm_xyz
sent: 2 min ago
preview: "hey, been a while — saw your post on the new ETHGlobal launch. quick coffee call? grabbing a few people on this btw → calendly.com/alex-real/30"
}
> cyphervex.analyze() {
pattern_match: DORMANT_REACTIVATION
confidence: 0.91
baseline_drift: last contact 11mo ago · prior cadence: monthly · this opener: unprecedented
tells:
· generic reference to a public event
· third-party-invite framing
· calendly link in opening message · scheduling pressure
· username variant of a recently-spoofed Paradigm associate
action: filter
recommend: verify identity through a known channel before any call.
}
Read in 1.4s — cost $0.003 — saved an estimated $40k
i

Per-counterparty memory.

Every read updates a behavioral baseline for the person on the other end. The twentieth read of someone is sharper than the first.

ii

Clinical framework.

Patterns are reviewed by a board-certified psychiatrist. Built on forensic deception-detection literature.

iii

On-chain outcomes.

Every flag is verifiable. Funds moved, or they stayed. The dataset trains on what happened — a label nobody else has.

003 — The roadmap

Consumer first. Infrastructure inevitable.

i
Now

Consumer Telegram filter.

Direct-to-user behavioral analysis for the first 1,000 onchain finance operators.

ii
Q3 2026

Protocol partnerships.

Aave, 1inch, MakerDAO sponsor free access for their active users as community safety.

iii
2027

Behavioral risk API.

Wallets and exchanges integrate to score counterparty risk on the conversation side.

iv
Beyond

Threat intelligence.

Custodians and crime-fighting units license the outcome-labeled dataset.

004 — Who's building this

Criminology meets crypto. A specific founder for a specific problem.

Jessica Salomon — Founder

Cambridge-trained criminologist. Onchain finance operator.

MA in Criminology from Cambridge. Fieldwork on psycholinguistic deception detection — the academic foundation of the Cyphervex framework. Honors BA from UC Berkeley.

First non-developer hire at MakerDAO. Growth at Aave and 1inch. Six years of onchain finance from genesis — the network turns into design partners on the first call.

MA Criminology · Cambridge
BA · UC Berkeley
Previously MakerDAO
Previously Aave
Previously 1inch
Dr. Tyler Vestal, MD — Clinical Advisor

The clinical eye on the framework.

MD from Columbia University. BA from USC. Board-certified psychiatrist.

Reviews the behavioral framework and grounds the pattern definitions in forensic deception-detection literature.

MD · Columbia University
BA · USC
Board-certified psychiatry
005 — Access

Fifty slots. Operators only.

Free during beta. For active onchain finance users with high inbound DM volume. Fifteen minutes of feedback a week. Tell us who you are.

Encrypted at rest — delete anytime