Apate says its decoy AI victims have kept scammers talking for 17 years

Divided out, the headline figure is small per conversation. Apate says its agents wasted 17 years of scammers' time across 2.5 million conversations ^[13]. That is about 8.94 million minutes, or roughly 3.6 minutes a conversation ^[11]. Dali Kaafar, the founder and CEO, kept the family-picnic scam call that gave him the idea going for 44 minutes ^[6]. That is about 12 times the average ^[12].

Brad Joffe, the co-founder and chief commercial officer, sells the product on timing. "Most of the industry works after the fact. Telcos block numbers, but scammers just get new ones. Banks watch transactions, but by then the scammer has already reached the customer. We go to the scammer directly and get the information before anyone gets hurt," Joffe told TechCrunch ^[5].

In our view a short average is fine. A scammer who hands over a wallet address in the second minute is worth more to a fraud team than one who chats for forty. TechCrunch's report gives totals for conversations and time but not how many conversations ended with a usable account, wallet or website, and that is the count a buyer needs.

The totals also come from the company itself. Apate is a finalist at the Startup Battlefield 200 at TechCrunch Disrupt and will pitch a panel of VC judges ^[9].

The loss data in the report shows why a bank would take the meeting. The FBI received more than 22,000 complaints of AI-related crimes in 2025, with about $893 million in losses ^[7]. More than $632 million of that, roughly 71%, came from investment scams ^[8][10]. In those, the FBI says, fraudsters use AI to run conversations and to deepfake the voices and appearances of trusted figures ^[8].

This is for fraud teams at institutions like Commonwealth Bank of Australia and TPG Telecom, both named as customers ^[14]. We would start a pilot with the mission feature ^[4], aimed at a question the team already has, and sort everything that comes back on two tests. The first is authority: whether the item concerns the team's own customers and brand, so it can act without anyone else. The second is shelf life: whether the item will still be valid when the team gets to it.

An item that passes both goes straight to action. One that is durable but belongs to another institution goes to a referral queue. One inside the team's authority that expires fast needs automation or gets dropped. The rest is noise. The pilot's score is the number of items in the first group.

The cost is analyst time. Someone has to do the sorting before any vendor total means anything for the team's own book of customers.