Apate sends AI agents undercover to gather intelligence on scammers

Turning conversations into intelligence

Founded in 2024 from research at Macquarie University in Sydney, Apate deploys thousands of agents that mimic people of different ages and genders, speaking with different accents and in different languages. They answer scam calls, respond to suspicious messages and emails, follow phishing links and join social media groups to start conversations.

The company says its agents talk with tens of thousands of scammers each day across phone calls, messages, email and the internet. Clients can direct agents to seek specific information, such as a bank account number or crypto wallet address. Text-based agents look for possible scams on forums, investment groups and social networks, then engage under fabricated identities.

Apate gives clients a portal with analytics and information about schemes targeting their users, employees, brands and products. Its customers include Commonwealth Bank of Australia and TPG Telecom, along with other utilities, telecom companies and government agencies.

The startup says it passes what it learns to clients, which use the information to develop anti-fraud measures. Its agents run on advanced models and Apate’s own fine-tuned models, trained on agent conversations with scammers and research by founder and CEO Dali Kafaar at Macquarie University.

From a family picnic to a company

The idea began with a scam call. Several years ago, Kafaar was at a family picnic when he answered an unfamiliar number. Realising it was a scammer, he played along for the sake of the children and kept the caller talking for 44 minutes.

The family laughed, but Kafaar saw a practical point: the scammer had spent that time talking to him instead of a potential victim. Kafaar had lost 44 minutes too, so he wondered whether the tactic could be automated—and whether the conversations could produce useful intelligence. That question became research with colleagues at Macquarie University, using machine learning and natural language processing to understand scam calls and improve fraud detection and prevention.

Co-founder and chief commercial officer Brad Joffe told TechCrunch that most companies respond after scammers have acted: telecom operators block numbers, only for criminals to get new ones, while banks monitor transactions after a customer has been contacted. Apate’s pitch is to engage scammers directly and collect information before someone is harmed.

The approach comes as AI gives scammers and hackers more ways to operate. In 2025, the FBI received more than 22,000 complaints about crimes involving AI, with losses of about $893 million. More than $632 million was tied to investment fraud. The FBI said criminals use AI to conduct conversations and impersonate trusted people, including celebrities and executives, by copying their voices and appearance.

People have long tried to waste scammers’ time, sometimes getting them to read entire books aloud or even have ridiculous tattoos made. Apate industrialises that impulse, with a different aim: to turn the interaction into evidence its customers can use.

The evidence behind the pitch

Joffe said TPG redirected nearly 590,000 calls to Apate’s agents, while Commonwealth Bank publicly reported fraud losses falling by 76%. An unnamed Australian bank used agents’ outbound calls to disrupt 22 scam campaigns. Those examples suggest that the product is not just a conversational novelty; clients are routing real calls to it and using its intelligence in fraud prevention.

But the announcement leaves a key measure unclear: how much of the intelligence gathered leads to prevented losses, rather than more information about scams? The reported outcomes are promising, but they are not directly comparable. A reduction in one bank’s losses, calls redirected by a telecom and campaigns disrupted by another bank each describe a different part of the process.

I think that gap is the central test for Apate. The company can point to millions of conversations and years of scammers’ time, but neither figure says how often an agent uncovers actionable information—or how reliably a customer can turn that information into a prevented fraud.

Apate raised $8.15 million in seed funding in August from Lobby Capital, OIF Ventures, Investible, Concept Ventures and Baobab Ventures. Joffe said the company has 20 employees and has raised just under $10 million in total. It plans to use the latest funding to develop its product and expand internationally, including opening an office in London and growing its presence in the US.

Apate is named after the Greek goddess of deception. Joffe said the company treats scammers swearing at its bots as a sign of frustration—and celebrates each instance. The joke captures the startup’s premise, but the harder measure of success is whether those conversations give customers enough warning to stop the next scam before it reaches a real person.