CLEVI, a startup attracting attention in Australia’s AI industry, is showcasing technology based on its proprietary large-scale generative AI model ‘Ivy’, featuring multimodal processing, real-time learning, agent models, security, and on-device and on-premises support. This technology is being evaluated as a foundation for expanding the scope of applications across the private and public sectors.
CLEVI’s Ivy model features multimodal capabilities that enable it to understand and process various types of data simultaneously, including text, images, voice and video. It also enables complex task automation and the provision of customised AI solutions through its code interpreter and expert agent creation capabilities. These outstanding technical capabilities stem from proprietary architectural innovations, including the distillation and synthesis of multiple small-scale models, and advances in attention structures and layer design.
In particular, CLEVI is paying close attention to security and privacy protection. It provides on-premises installation options for corporate and public-sector customers and requires only the minimum amount of information when users sign up for its services, pursuing AI services based on trust.
Having launched Ivy’s open beta service in March 2025, CLEVI is accelerating its market entry with high-performance, low-cost AI solutions. Industry observers said, “CLEVI is one of the few companies in Australia to possess its own foundation model and is highly regarded for both its technological capabilities and market adaptability,” adding that “its strategy of expanding its business beyond the private sector into the public sector will lead to global competitiveness in the future.”
Photo source = Industrial AI EXPO
CLEVI plans to continue growing into a leading AI company representing Australia through innovation that pushes beyond the limits of AI technology and customised solutions tailored to various industries.
Seongjun Kim, reporter ([email protected])
[Copyright ⓒ Digital Times. Unauthorised reproduction and redistribution, and use for AI training and applications, are prohibited]
Source: Digital Times (Reporter Kim Seong-jun) | https://www.dt.co.kr/article/12031314?ref=naver