Aviel Intelligence: The Startup Using AI "Honeybots" to Catch Scammers

Most fraud prevention software watches for suspicious transactions after the fact.

Aviel Intelligence built something stranger:

AI personas designed to actively get targeted by scammers.

The London-based startup calls them honeybots. Synthetic personas that pose as real people online, engage scammers in conversation, and extract the bank account details scammers ask victims to pay into.

Founded in 2024 by Joe Tallett, Peter Griffin, and a small team of former Google employees and cybersecurity specialists, Aviel raised £350,000 in pre-seed funding led by Fuel Ventures, with Cambridge Angels and other angel investors also participating.

Click here to see how two repeat founders are automating bank compliance with AI

Why This Specific Problem, Right Now

The UK has a genuinely large and growing scam problem.

Authorised push payment fraud, where victims are tricked into willingly sending money to a scammer, reached £257.5 million in losses in just the first half of 2025 alone, a 12% year-over-year increase.

The regulatory landscape shifted meaningfully in October 2024.

The UK's Payment Systems Regulator now mandates that financial institutions reimburse APP fraud victims, moving the financial liability directly onto banks and payment service providers rather than leaving it with individual consumers.

That regulatory shift is central to why Aviel exists. Once banks are on the hook financially for fraud losses, upstream, intelligence-led prevention tools become a lot more commercially valuable than they were when consumers bore the loss themselves.

How the Honeybots Actually Work

A honeybot is a fake persona, an AI-generated identity designed to look and behave like a real, engageable target online. Scammers approach it the same way they'd approach any real potential victim.

As the scammer works the honeybot toward a payment, they eventually provide a bank account to send money to.

That account gets flagged and fed directly into Aviel's real-time intelligence stream to partner banks.

Griffin has described one real, specific example: a scammer targeting World Cup ticket buyers sent a honeybot a fake ticket offer and a bank account to pay into.

The account was flagged and disabled by Aviel's banking partner before any real victim could be scammed through it.

Griffin has said Aviel captures this kind of mule account intelligence at a scale no human investigative team could realistically replicate, since honeybots can run these conversations continuously and in parallel across scam networks.

Here's the startup whose product accidentally crashed insurance stocks by 9% in a day

What Makes This Different From Standard Fraud Tools

Most fraud detection software is reactive, watching transaction patterns for anomalies after money has already started moving. Aviel's approach is explicitly upstream, aiming to identify and disable mule accounts before a real victim's payment ever reaches them.

Tallett has framed the company's mission around a specific mechanism: extracting operational intelligence directly from scammers themselves, rather than only inferring fraud patterns from transaction data after the fact.

The company has also built what it calls the EVIE engine, its proprietary system for running and coordinating honeybot personas and turning scammer conversations into structured, actionable intelligence banks can act on quickly.

Early Traction and Recognition

Aviel is genuinely early stage, still operating on a modest pre-seed round. But it's already picking up meaningful industry recognition relative to its size.

The company was named a finalist across multiple categories at a Money20/20 Europe 2026 startup competition, including Best Financial Crime Prevention Initiative and Best Use of AI in Payments, competing against better-funded rivals in the same space.

Aviel has also formed a relationship with TSB Bank Labs, giving it direct access to a real banking partner for testing and deploying its intelligence feed in production, a meaningful validation step for a company this young in a space where bank trust is genuinely hard to earn.

See how one founder is insuring the physical buildings powering the AI boom

The Honest Caveat

Aviel's approach depends on scammers continuing to fall for synthetic personas at scale, which is inherently an arms race rather than a permanent solution.

As AI-generated personas become a known tactic in the fraud-prevention industry, sophisticated criminal networks will likely adapt their own screening methods to detect and avoid honeybots over time.

That doesn't make the current approach ineffective, but it does mean Aviel's long-term durability depends on continuously staying ahead of adversaries who are themselves increasingly using AI to scale their own operations.

Griffin has acknowledged this dynamic directly, describing the fraud landscape as an industrialization of scams matched by an industrialization of countermeasures, which is an honest framing of an ongoing race rather than a solved problem.