AI startup Memories.AI has secured $8 million in seed funding to further develop its platform aimed at enhancing video understanding through long-context memory. The round was led by Investors Fund, Samsung Next, LDV Partners, First Spark Ventures, Seedcamp, and Creator Ventures, and marks a significant step toward creating AI systems that can retain and process thousands of hours of video content.
Founded in 2023 by former Meta AI leaders George Livadas and Dongjoon Hyun, Memories.AI is building a memory infrastructure specifically for AI systems that require persistent and contextual understanding of video data. Unlike traditional AI models that rely on short bursts of information, Memories.AI enables persistent video memory—allowing AI agents to “remember” what they’ve seen and learned over extended timeframes.
The Problem: AI's Short-Term Memory in Video Analysis
Most current generative AI tools, including those built on large language models (LLMs), operate within a constrained context window—often limited to a few thousand tokens or several minutes of video. This short memory creates significant challenges when applying AI to surveillance, sports analysis, long-form content editing, or any domain requiring deep temporal understanding.
Memories.AI addresses this gap by developing a system that allows AI to retain key visual and audio cues over long durations, which could dramatically improve applications like film editing, security footage analysis, and virtual assistant memory.
Persistent Memory for AI Agents: A New Paradigm
The company’s vision aligns with the growing demand for AI agents that can interact, adapt, and improve over time. By enabling long-context video memory, Memories.AI supports use cases where an AI agent needs to recall past conversations, visual patterns, or behaviors that occurred hours, days, or even weeks earlier.
This evolution is especially critical for emerging multimodal systems combining text, audio, and video. With its infrastructure, Memories.AI aims to become a backbone for a new wave of persistent, intelligent agents—spanning from customer support to autonomous video creators.
Strategic Backing from Samsung and Other Global Investors
Samsung Next’s involvement adds strategic weight to the company’s roadmap, hinting at future collaborations in consumer electronics, surveillance, or smart assistant ecosystems. Other investors, like LDV Partners and First Spark Ventures, bring expertise in AI, media, and hardware convergence.
The funding will be used to expand the technical team, scale memory infrastructure, and move the product from prototype to broader deployment. While still in early stages, Memories.AI is positioning itself as a foundational layer for persistent AI across enterprise, creative, and personal use cases.
Building Toward Smarter, More Context-Aware AI
George Livadas, co-founder and CEO, stated in a release, “AI models today forget too much, too quickly. We're building the memory layer so AI can understand and remember the world the way people do.”
As AI adoption in video grows—from content moderation to intelligent editing suites—Memories.AI's long-context model could be a transformative enabler. The company joins a rising cohort of startups focused on enhancing AI cognition, particularly in time-rich, multimodal environments.