AI for Video Content: Multimodal Agents with Meenal Nalwaya of Reka

Reka's Multimodal AI Approach

• 💡 Reka focuses on building enterprise-standard multimodal agents that can process and reason across visual and audio data, not just text.
• 🎯 Their differentiation lies in being multimodal-first, addressing the vast amount of untapped visual and audio data, and being a full-stack company that builds its own foundational models.
• 🔑 This full-stack approach provides control over quality, customization, and development speed, avoiding reliance on third-party roadmaps.

Key Platform Capabilities: Recision and Research

• 👁️ Recision enables human-like understanding of images and videos, allowing users to search and ask questions directly about visual content.
• 📈 Examples include generating highlight reels from sports footage or identifying security events from surveillance feeds.
• 🧠 Reka Research adds a reasoning layer, enabling deep research across private enterprise data and the web, providing structured and cited answers.
• 📊 This combination of perception (Recision) and reasoning (Research) forms the building blocks for advanced multimodal agents.

Real-World Applications and Impact

• 🚓 In security, Reka solutions with the Ohio Police Department led to a 65% faster case resolution and a 42% reduction in crime, saving officers significant time.
• 🎬 For media companies like Paramount, Reka unlocks value from vast video archives by making content searchable at granular levels (scene, theme, actor, emotion), improving licensing and discovery.
• 📺 This enhanced discovery can lead to more consumer engagement and higher licensing values, transforming how content is found and consumed on platforms like Netflix.
• 📰 News media can repurpose existing content into new formats quickly for social media using Reka's reel generation, increasing audience engagement without additional production costs.
• 💰 Publishers can also increase advertising and sponsorship revenue by proving ROI through logo appearance tracking and selling premium ad placements during moments of excitement.

Future of AI and Advice for PMs

• 🚀 The AI landscape is shifting from text-centric to multimodal, with images, video, and audio becoming increasingly important data sources.
• 🏢 A contrarian view suggests a move towards full-stack AI companies for greater control over quality, reliability, and pricing as complex use cases emerge.
• 🎯 For Product Managers in AI, the advice is to anchor on customer problems and leverage evaluations as the new KPIs to ensure products solve real pain points and are built in the right direction.
• 💡 PMs remain crucial for defining the 'what' and 'why', even with advanced coding tools, making it an exciting time for the role.