Siblings Patrick and Ryan Coughlin, each boasting notable careers in technology (Patrick’s background includes national cyber defense, Splunk, and Cisco while Ryan has worked on consumer products at Apple and Spotify), have introduced a new type of security startup. 

Savi Security aims to shield everyday individuals from the latest wave of remarkably convincing AI-generated scams, whether these are delivered via text messages, emails, or phone calls. 

The company recently secured $7 million in seed funding and is set to launch its app for both iPhone and Android on Tuesday. This funding round was spearheaded by Acrew Capital, with support from Magnify Ventures, TTCER, and Resolute Ventures. 

The founding inspiration for the company stemmed from a terrifying experience involving their mother.  

Approximately two years ago, Patrick Coughlin received an upsetting call from his mom, who reported that she had been contacted by a man claiming he had abducted Coughlin’s sister. At that time, he was serving as senior vice president of security products at Cisco, having joined the company following the acquisition of his cloud security startup TruSTAR by Splunk for an estimated $82 million in May 2021. In 2024, Cisco went on to acquire Splunk.

Coughlin remembered that her mobile phone displayed his sister’s caller ID. During the conversation, “she believes she hears my sister’s voice pleading, ‘Mom, they’ve got me.’ Then there’s a horrifying scream, followed by my sister saying, ‘You need to do what they say.’ Soon after, a man comes on the line and states, ‘If you don’t send us $1,200 immediately, we will kill your daughter in the nearby Walmart’s parking lot,’” he recounted. 

The scammer had expertly spoofed Coughlin’s sister’s number, mimicked her voice, and mentioned the local Walmart she frequently visited. 

Fortunately, the mother stayed calm, called her daughter, and verified that she was safe. The kidnapping was nothing more than an AI-generated scam.  

Coughlin, much like his mother, was deeply unsettled. 

“After calming my mom down, I found myself thinking: What has fundamentally altered in the underlying cybercriminal landscape that now allows us to leverage the same sophistication previously directed at government entities, and later at Fortune 500 companies? And now we’re seeing that sophistication aimed at everyday consumers?”  

The answer is, of course, inexpensive and potent large language models (LLMs) and other generative AI tools. 

Prior to the advent of AI, targeting consumers for such scams was not financially viable. It necessitated extensive research on the victim, technology for voice spoofing, and similar resources. Such scams were mainly directed at individuals with substantial wealth, such as corporations or governments, as was the technology needed for their defense.

“There’s a shift happening now regarding consumers and AI within the hands of cybercriminals,” Coughlin explains. The expenses involved in conducting these scams have diminished significantly, and the necessary research materials are readily accessible. 

“You can replicate a voice from merely three seconds of audio taken from a publicly available social media post. We all hold traces of content out there in the ether — like casual conversations or narrating a kid’s football game while recording it for Facebook.” 

The FTC reported last month that victims of online fraud collectively lost $3.5 billion to impostor scams in 2025, three times the losses reported in 2020. While older Americans make up the majority of those reporting such scams, some studies suggest that Gen Z is also particularly vulnerable. Research from 2025 conducted by Malwarebytes, a provider of antivirus and anti-malware solutions, indicated that Gen Z individuals encountered text scams more frequently than other generations, falling for them approximately 25% of the time. 

The Coughlin brothers aimed to create an immediate intervention tool. 

They tested their concept, along with the AI scam detection model they were developing, by launching a free platform named Scam Wise. It requires no registration, allowing users to anonymously upload any suspicious texts, images, or emails, and Scam Wise will ascertain if they are likely fraudulent. 

“We rolled that out around four months ago. We’ve received 50,000 submissions so far, and this number grows by approximately 10,000 submissions each week,” Coughlin stated. 

Scam Wise has provided a valuable source of real-world data to enhance Savi’s scam detection AI model. Currently, the startup mainly utilizes Google’s Gemini, but has constructed its software on an AI gateway to leverage additional AI models as necessary, such as those specifically targeting voice detection. 

On Tuesday, Savi introduced a paid product, an app for iOS and Android designed for consumers that can assess texts, voicemails, and incoming calls for potential scams.

Although such features can be found in various products (such as Malwarebytes), Savi’s standout feature is its live call monitoring capability. 

During a suspicious phone call, a user can opt to have the app’s live agent listen in. Savi monitors for behavioral cues that could indicate fraudulent activity while the call is ongoing. 

Savi’s pricing is somewhat unconventional. It charges $8/month, discounted to $63/year, covering an entire family, and imposes no limit on the number of users. A single subscription can encompass a person’s children, spouse, parents, and anyone else the main account holder wishes to add for administrative assistance. 

AI has transformed the accessibility for “becoming a fraudster,” Coughlin stated. “We’re facilitating the entrance into fraud due to the diminished barriers to deceiving individuals. Consequently, we not only face organized criminals and syndicates, but also everyday individuals being lured into committing fraud.” 

Savi Security’s solution resembles a new generation of anti-virus-like software: one that employs AI in real-time, mirroring the methods utilized by fraudsters.