Aviel Intelligence
Pre-seed London startup that deploys AI 'honeybot' personas to engage scammers directly and feed banks real-time mule-account intelligence.
Visit Website ↗ + Add to Compare Claim This CompanyOverview
Aviel Intelligence takes an unusually direct approach to fraud prevention: rather than only analyzing transaction data after the fact, it deploys AI-generated synthetic personas — “honeybots” — that engage scammers in conversation across messaging platforms, including as conversations move into encrypted channels like WhatsApp. As scammers interact with these personas, Aviel extracts details about the mule accounts and infrastructure being used, then feeds that intelligence to banks and payment service providers as a live feed so they can flag suspicious accounts before money moves.
Founded in London in 2024 and led by CEO Joe Tallett, alongside co-founders with backgrounds including Google and what3words, Aviel raised £350,000 in pre-seed funding in April 2025, led by Fuel Ventures with participation from the Cambridge Angels. The company has since won the Start-Up Pitch Competition at Money20/20 Europe 2026, and its intelligence engine is live and operating, though it remains a very early-stage company without disclosed enterprise-scale deployment data.
Innovation Matrix Assessment
Launched in 2024 and already has a live, operating intelligence engine and a competition win (Money20/20 Europe 2026) less than two years in — reasonable pace for a team of this size.
If accurate, real-time mule-account intelligence could let banks and PSPs intercept scam-related transfers before funds move, but with no disclosed customers yet, operational impact is currently unproven rather than demonstrated.
Momentum is genuinely early-stage: a single £350,000 pre-seed round and a pitch-competition win are the only independently verifiable signals; no named bank or PSP customers or deployment scale were disclosed.
Proactively engaging scammers through AI personas to extract mule-account details is a different approach from typical transaction-pattern monitoring, though it remains an early, unproven technique rather than a demonstrated shift in how fraud is stopped.
Aviel states its intelligence engine is 'live' and identifying scam networks, but no independent verification, named bank partnerships, or fraud-losses-prevented figures were found.
Authorized push payment fraud and scam losses are a well-documented, growing problem for UK and EU banks, so the underlying need is durable even though this specific technique is unproven at scale.
Why CISOs Should Care
Gives fraud and security teams at banks/PSPs an earlier warning signal — mule-account intelligence gathered directly from scammers — before a fraudulent transfer completes.
What Makes It Different
Uses AI-generated personas to actively engage and extract intelligence from scammers, rather than relying solely on post-hoc transaction-pattern analysis.
The Matrix Verdict
47/100 — EMERGING / UNRANKED
Emerging/Unranked, worth watching. Aviel's honeybot approach to fraud intelligence is a creative, verifiably different technique, but at pre-seed stage with no disclosed bank customers or independently confirmed results, it registers as an early, unproven bet rather than a validated innovator.
Editorial Note: Claims vs. Verified Findings
Claims that the intelligence engine is 'live' and actively identifying scam networks come from Aviel's own materials; no independent case study or named financial-institution partner was found to corroborate results.
Sources
Alternatives to Aviel Intelligence
Unknown Cyber Inc.
CISO ReviewedMalware genomics platform using automated deep static analysis and code-lineage comparison to identify unseen malware, variants, and supply-chain…
Synack Inc
A penetration-testing-as-a-service platform pairing a vetted researcher community with AI-driven attack surface discovery for continuous security validation.
Horizon3.ai
Autonomous penetration testing company whose NodeZero platform self-attacks networks without persistent agents, aiming to replace annual manual pentests…
Airlock Digital
Application allowlisting (deny-by-default) platform that blocks unapproved executables, scripts, and processes to prevent ransomware and malware execution.
Cogent Security
Agentic AI platform that autonomously triages, investigates, and remediates vulnerabilities as a force multiplier for security teams.
Reality Defender
Deepfake and synthetic media detection company offering real-time detection across voice, video, image, and text for enterprises and…