The key facts

  • Apate's agents have held 2.5 million conversations and, according to the company, have collectively taken 17 years of scammers' time
  • Clients include Commonwealth Bank of Australia and TPG Telecom; TPG has redirected nearly 590,000 calls to the agents, and Commonwealth reports a 76% drop in publicly reported scam losses
  • In August the startup raised $8.15 million in seed funding; it has raised just under $10 million in total

AI agents instead of victims

Australian startup Apate.ai, founded in 2024 out of research at Macquarie University in Sydney, has deployed thousands of AI agents that disguise themselves as ordinary people and deliberately engage with scammers. The agents answer scam calls, correspond via email and messengers, click phishing links and join social media groups. They portray people of different ages, genders and accents and speak different languages, keeping scammers on the hook for as long as possible.

What this approach delivers

According to the company, its agents have held 2.5 million conversations and collectively taken 17 years of scammers' time. Apate's clients — banks, telecom companies, government agencies and utilities — use data that is usually only obtained after a victim has already been defrauded: bank account numbers, crypto wallets, phone numbers, malicious websites, the brands scammers impersonate, and other details of the schemes. Co-founder and chief commercial officer Brad Joffe told TechCrunch that the agents talk with tens of thousands of scammers daily via calls, messages, email and the internet, and clients receive the results of these conversations. The company uses both frontier models and its own fine-tuned models, trained on the agents' conversations and on Kaafar's research at Macquarie.

Clients and results

Among the startup's clients are Commonwealth Bank of Australia and TPG Telecom, as well as other telecoms, utilities and government agencies. According to Joffe, TPG has redirected nearly 590,000 calls to Apate's agents, Commonwealth has recorded a 76% reduction in publicly reported scam losses, and one unnamed Australian bank used outbound agents to disrupt 22 fraud campaigns. Clients also get a portal with analytics on scam activity and can assign agents specific tasks, such as obtaining an account number or a crypto wallet address.

Origin of the idea

The idea came from a real scam call: founder and CEO Dali Kaafar, at a family picnic several years ago, decided to 'play naive' for the sake of his children and kept a scammer on the line for 44 minutes. He noticed that this was time the scammer did not spend on a real victim, and began thinking about automating the process. Later, at Macquarie University, he and colleagues applied machine learning and natural language processing to analyze the content of scam calls and develop methods for detecting them.

Context and funding

Apate is one of the finalists of Startup Battlefield 200 at TechCrunch Disrupt, which took place in San Francisco through Thursday, October 15. According to TechCrunch, citing the FBI, in 2025 the agency received more than 22,000 complaints about crimes involving AI, with losses of about $893 million, of which more than $632 million came from investment fraud using AI-generated dialogue and deepfakes. In August, Apate raised an $8.15 million seed round from Lobby Capital, OIF Ventures, Investible, Concept Ventures and Baobab Ventures; in total, the 20-person company has raised just under $10 million and plans to spend the funds on product development and international expansion, including an office in London and strengthening its presence in the US. The name references Apate, the Greek goddess of deceit. 'We are probably one of the few companies in the world that celebrates when our product gets sworn at,' Joffe said, noting that the company tracks scammers' cursing at the bots as a sign of frustration.