- Nexusflow asserts that in security applications, NexusRaven attains an impressive accuracy rate of up to 95%, surpassing OpenAI LP’s GPT-4, which achieves 64%.
- Nexflus Copilot, a feature of the company’s offering, provides a cohesive conversational interface that translates instructions into actionable insights.
Nexusflow.ai Inc., a cybersecurity start-up using generative AI, recently disclosed that it had raised USD 10.6 million in new funding in a round co-led by Point72 Ventures LP, Fusion Fund LLC, and several individual investors.
Professors Jiantao Jiao and Kurt Keutzer from the Berkeley AI Research Lab and Jian Zhang, who earned his doctorate from the Stanford AI Lab and previously held the position of Director of machine learning at SambaNova Systems Inc., founded Nexusflow. According to the company, it uses generative AI to maximize the capabilities of security products and enhance security operations and engineering.
Businesses can own AI models using its NexusRaven generative AI platform without risking the exposure of confidential information. According to Nexusflow, NexusRaven’s accuracy rate in security applications can reach up to 95%, compared to 64% for OpenAI LP’s GPT-4.
NexusRaven’s 95% accuracy enables businesses to combine fragmented data from numerous software tools and knowledge sources, speeding up team decision-making and streamlining security operations workflows. Additionally, it is claimed that NexusRaven-13B will outperform GPT-4 regarding cost savings and inference speed.
In a zero-shot setting, NexusRaven-13B outperforms all other comparable-sized open-source large language models by generalizing to tools that were not used during model training and achieving a success rate comparable to GPT-3.5. Users have complete control over the model when used in commercial applications because NexusRaven-13B was trained without using any data produced by exclusive LLMs like GPT-4.
Nexflus Copilot, a different aspect of the company’s product line, provides a unified conversational interface that transforms commands into action insights. A command function in Copilot enables users to use security tools without having to undergo complex learning curves, and it can synthesize fragmented data to hasten decision-making.
Jiao, Chief Executive Officer, said in a statement, “Nexusflow specializes in curating high-quality data for data-centric AI. We also leverage our expertise in generative AI systems to ensure efficient implementation and scalability, resulting in significantly lower costs compared to OpenAI’s GPT-4.”
Although it’s unclear if Nexusflow has raised money before, a leading media house claims the company raised new fundings of USD 53 million. By the end of the year, the company plans to hire 20 people, up from its current 10 employees.