Перейти к содержимому

Вывод: Инференс, безопасность, данные, предприятие

Grace Gong

0:00 / 0:00

Вывод: Инференс, безопасность, данные, предприятие

20 просмотров · 2 недели назад
Grace Gong
8,75 тыс. подписчиков
20 просмотров · 2 недели назад
Where is enterprise AI heading as inference scales, open models improve, and AI agents gain access to real-world systems and data? In this panel, Kai Mak, CRO of Together AI; Nancy Wang, CTO of 1Password; Ross Fubini, Managing Partner at XYZ; Aidan Madigan-Curtis, Partner at Eclipse; and Jordan Tigani, CEO of MotherDuck, join Grace Gong for a wide-ranging discussion on inference, security, data, open-source models, and enterprise AI. The conversation explores what is actually driving AI inference demand, from software agents and enterprise workloads to customer service and the emerging wave of personal agents. The panel also examines why open-source models are becoming increasingly important as companies look for better economics, greater data control, and the ability to run models locally or on the edge. Security becomes even more critical as AI agents move from generating answers to taking actions. Nancy Wang explains why agents need secure access to enterprise systems, how authorization may evolve toward models reasoning about the risk of individual requests, and why AI will increasingly be needed to defend against AI-powered attacks. The speakers also discuss the changing role of data and context as companies use multiple models for different workloads, why strong evaluations become more important in a multi-model environment, and how enterprises can balance model performance, cost, privacy, and security. The panel also digs into token economics and AI adoption, including when companies should use frontier models versus smaller or open models, why token consumption is becoming an executive-level concern, and how companies can route workloads across models based on the desired outcome. Finally, the conversation tackles a bigger question for the AI industry: will model providers and application companies increasingly compete with one another? The panel explores the potential collision between model companies and breakout applications, the importance of proprietary data, and why companies building today may need to think carefully about how to stay on top of the AI wave as models, infrastructure, and applications continue to converge. Panelists: Kai Mak, CRO, Together AI Nancy Wang, CTO, 1Password Ross Fubini, Managing Partner, XYZ Aidan Madigan-Curtis, Partner, Eclipse Jordan Tigani, CEO, MotherDuck Moderator: Grace Gong, Founder & CEO, Smart Venture Media #AI #ArtificialIntelligence #EnterpriseAI #OpenSourceAI #AIInfrastructure #Inference #AIAgents #Cybersecurity #Data #AIInvesting #VentureCapital #FutureOfAI