The CounterFlow AI founding team. From left: Erik Breuhaus, Randy Caldejon, Andrew Fast. Not pictured: Peter Shaw. Credit: Credit: Josh Mandell, Charlottesville Tomorrow
Crozet cybersecurity startup CounterFlow AI announced last week that it raised $2.7 million in seed funding from Osage University Partners, the Charlottesville Angel Network, and several individual investors.

“We are looking to grow and stay in Crozet,” said Randy Caldejon, co-founder and CEO of CounterFlow AI. “Our intent is for this to fund the true development of our machine-learning platform — to build the product and get it to market.”

Caldejon and Peter Shaw, CounterFlow AI’s COO, previously co-founded nPulse Technologies, a network security firm that was acquired by FireEye in 2014.“[Caldejon and Shaw] understand the market very well and have proved themselves capable of successfully starting up and selling a business,” said Craig Redinger, founder of the Charlottesville Angel Network.

Redinger said CounterFlow AI’s pitch convinced more than 15 members of the Charlottesville Angel Network to contribute a total of $600,000 in funding — one of the group’s largest investments to date.

CounterFlow AI hopes to help cybersecurity analysts use artificial intelligence to identify, evaluate and eliminate threats to their networks. 

Previously, cybersecurity firms have tried to establish a “perimeter” to completely block hackers from compromising networks — a goal now recognized as a lost cause, according to Andrew Fast, co-founder and chief data scientist at CounterFlow AI.

“That perimeter always is a leaky fence,” Fast said. “Threats still keep getting through. … Pretty much every business has been hacked, or will be hacked.”

Co-founder and CFO Erik Breuhaus said businesses cannot find enough qualified cybersecurity analysts to handle the growing volume and sophistication of online threats. Many businesses, governments and other organizations are turning to artificial intelligence to fend off hackers. 

CounterFlow AI has developed network security sensors that incorporate machine learning, a branch of artificial intelligence that enables computers to learn from data without human intervention. 

Perrone Robotics, another software startup in Crozet, uses machine learning to guide self-driving vehicles. CounterFlow AI is using it to find irregularities in network activity and differentiate false alarms from true cyberattacks. 

Breuhaus said it still can take weeks for businesses using standard machine-learning software to realize they have been hacked.

“It can take days to analyze data with machine learning, after many days of waiting for that data to pool up,” Breuhaus said. 

CounterFlow AI aims to speed up threat detection with “streaming” machine learning, which takes in a constant flow of real-time data from a network.

The company is testing its software on network security devices that can transfer up to 100 gigabits of information per second. 

“The real-time detection is the critical differentiator [for our platform],” Breuhaus said. 

“Machine learning usually is focused on data at rest,” said Fast. “Streaming machine learning looks at data in motion.”

Caldejon said CounterFlow AI’s threat-hunting software will be named after one of the most agile and prolific hunters in the animal kingdom: the dragonfly.

Neil MacDonald, a vice president and analyst studying cybersecurity for Gartner Research, said some global companies, such as DarkTrace and ProtectWise, also claim to offer real-time threat detection. “It is pretty cutting-edge,” he said.

However, MacDonald said, simply identifying threats isn’t enough to defeat them. “Detecting is great, but you also need to respond,” he said.

Breuhaus said the CounterFlow AI platform can work in concert with automated firewalls and security gateways, and provide analysts with contextual information about a threat.

John Lee, a principal at Osage University Partners who oversaw the Philadelphia-based fund’s investment in CounterFlow AI, said reliable security solutions still require human expertise. 

“Putting a human in the loop and empowering the analyst to monitor their network at scale is the right place [for CounterFlow AI] to be in,” Lee said. 

Fast said CounterFlow AI’s ease of use and integration with other products also will help the product stand out from competitors. “There is currently a gap between what computer scientists can do [with machine learning] and what analysts can handle and use,” he said.

Breuhaus said CounterFlow AI expects to use its seed funding to make several strategic hires this year. While the company’s software developers could be based remotely, he said its new data scientists likely will work at the Crozet headquarters.

“We are excited about the growth of the data science ecosystem in Charlottesville,” Breuhaus said. 



Josh Mandell graduated from Yale in 2016 and has been recognized by the Virginia Press Association with five awards for education writing, health, science and environmental writing and multimedia reporting.