Australian startup Apate.ai runs thousands of AI agents whose whole job is to get scammed. They pose as ordinary people, answer scam calls, reply to phishing emails and keep fraudsters talking long enough to pull out useful intelligence. TechCrunch AI reports that the company is one of the finalists at this year’s Startup Battlefield 200 at TechCrunch Disrupt in San Francisco, which wraps up today.
Most scam-fighting tools wait until a victim gets hit. Apate tries to get the information first. Here’s how it works and why banks and telcos are paying for it.
An army of fake victims
Apate’s agents play people of different ages, genders and accents, and they speak several languages. They answer calls, reply to dodgy messages, click phishing links and join social media groups to draw scammers in. According to the company, its fleet has wasted 17 years of scammers’ time across 2.5 million conversations. Co-founder and chief commercial officer Brad Joffe told TechCrunch the agents talk to “tens of thousands of scammers every day across calls, messages, email, and the web.”
Wasting time is the bonus. The intel is the product
Apate’s real pitch is that its agents get scammers to reveal details that usually surface only after someone has lost money:
- Bank accounts and crypto wallets used to collect stolen funds
- Phone numbers and malicious websites
- Which brands the scammers are impersonating
- How the scam itself works
Customers use this data to shape their anti-fraud work. “Most of the industry works after the fact,” Joffe said. “Telcos block numbers, but scammers just get new ones. Banks watch transactions, but by then the scammer has already reached the customer.”
Customers can send agents on missions
Customers don’t just get raw data. They get an insights portal with analytics on scams aimed at their users, employees, brands and products. They can also give agents specific goals, like getting a particular bank account number or wallet address. Apate’s text agents search forums, investment groups and social platforms for scam activity, then step into conversations using made-up identities.
Built on scam data and academic research
Apate runs on a mix of frontier models and its own fine-tuned models. It trains them on its agents’ past conversations with scammers and on research founder and CEO Dali Kaafar did at Macquarie University in Sydney. That research used machine learning and natural language processing to break down what happens on scam calls.
The idea started at a family picnic. Kaafar got a scam call, played dumb to entertain his kids and kept the scammer on the line for 44 minutes. He realized that was 44 minutes the scammer didn’t spend on a real victim. He also realized it was 44 minutes of his own time, so he started wondering if he could automate it.
Early customer results look strong
Apate’s customers include Commonwealth Bank of Australia, TPG Telecom, and several utilities, telcos and government agencies. Joffe shared these numbers:
- TPG has sent nearly 590,000 calls to Apate’s agents
- Commonwealth Bank has seen a 76% drop in publicly reported scam losses
- An unnamed Australian bank used Apate’s outbound agents to disrupt 22 scam campaigns
One caveat: these figures come from the company, and the 76% drop at Commonwealth Bank likely has more than one cause. Big banks run plenty of fraud programs at the same time.
Funding and expansion plans
Apate raised an $8.15 million seed round in August from Lobby Capital, OIF Ventures, Investible, Concept Ventures and Baobab Ventures. The 20-person company has raised just under $10 million in total. It plans to spend the new money on product work and international growth, including a London office and a bigger presence in the U.S. Apate hasn’t shared pricing. It sells to large institutions, not consumers.
Why this matters
AI has handed scammers some powerful tools. The FBI received more than 22,000 complaints about AI-related crimes in 2025, with about $893 million in losses. More than $632 million of that came from investment scams, where fraudsters use AI to run conversations and deepfake the voices and faces of celebrities and executives.
What stands out to me is the reversal. Scam-baiting has been an internet hobby for years, with people getting scammers to read whole books aloud. Apate takes that idea, scales it up with AI and sells the results to companies. If scammers can use AI to run thousands of conversations at once, defenders can do the same thing back.
The obvious next question is what happens when scammers start using bots to screen out bots. That arms race is probably coming. For now, Apate measures success in an unusual way. Joffe says the team celebrates every time a frustrated scammer swears at one of its agents. It’s fitting for a company named after Apate, the Greek goddess of deception.
You can find the full story at TechCrunch AI.