Insider Brief
- Graphon AI has emerged from stealth with $8.3 million in seed funding led by Novera Ventures, with participation from Perplexity Fund, Samsung Next, Hitachi Ventures, and others, to build what it describes as a pre-model intelligence layer that sits between raw enterprise data and AI foundation models.
- Rather than working within a model’s context window, Graphon automatically discovers relationships across multimodal data including video, audio, documents, images, and structured databases before the model ever processes it, enabling AI systems to reason over effectively unlimited and persistent context rather than isolated data fragments.
- Founded by engineers and researchers from Amazon, Meta, Google, Apple, and NVIDIA, Graphon is already deployed with enterprise customers including GS Group, one of South Korea’s largest conglomerates, where it has been used for applications including customer movement analysis in convenience stores and safety monitoring at construction sites.
PRESS RELEASE — Graphon AI emerged from stealth with $8.3 million in seed funding to build a new class of AI infrastructure: a pre-model intelligence layer that captures how data connects — making foundation models more accurate and capable of reasoning over unlimited multimodal data. The round was led by Arvind Gupta of Novera Ventures, with participation from Perplexity Fund, Samsung Next, GS Futures, Hitachi Ventures, Gaia Ventures, B37 Ventures, and Aurum Partners (the investment fund affiliated with the owners of the San Francisco 49ers and other strategic LPs).
Today’s AI models are powerful, but they all run into the same problem: they can only read a small chunk of data at a time. Even the most advanced LLMs are limited to roughly 1 million tokens, while organizations hold trillions of tokens across documents, videos, logs, and databases. Retrieval systems like Retrieval-Augmented Generation (RAG) can surface relevant content, but can’t discover hidden relationships across datasets. That leaves AI capable of answering discrete questions about individual pieces of data, but unable to reason about how that data connects.
Rather than working inside a model’s context window, Graphon operates before the model ever sees the data. Using graphon functions, the system automatically discovers how data connects across video, audio, documents, images, and structured databases. It can also analyze real-world data sources generated by devices and systems, from enterprise software, cameras, machines, smartphones, and emerging platforms like smart glasses. The result is a system that can handle effectively unlimited and persistent context and can work with any foundation model or agent framework, enabling new AI applications while making existing ones far more efficient.
“Graphon changes where the intelligence happens,” said Arvind Gupta, Founder and Managing Director at Novera Ventures, who made Graphon the first investment from the flagship fund. “Most companies are trying to build ever-larger models. Graphon is improving the layer between raw enterprise data and the model itself. That gives today’s foundation models a much better understanding of complex data — and makes them far more capable without needing to be bigger.”
Graphon’s founders and advisors bring together decades of experience in robotics, AI research and the application of graphons to real-world systems. The leadership team includes former researchers and engineers from Amazon, Meta, MIT, Rivian, Google, Apple, NVIDIA, Samsung AI Center, and NASA.
“AI has spent the last decade learning to mimic language,” said Graphon Founder and CEO Arbaaz Khan. “But the world isn’t made of tokens, it’s made of relationships. By preserving that structure, we make foundation models more accurate and more useful at enterprise scale. An LLM with Graphon is better than an LLM alone. We’re not replacing models — we’re amplifying them.”
Graphon customers and developers on the platform have already used Graphon for:
- Enterprise content management: Reasoning across video, audio, images, and documents simultaneously.
- Industrial intelligence: Spot process gaps, compliance issues and root-cause events across video plus enterprise systems.
- Agentic workflows: Let AI agents act on rich multimodal inputs to automate decisions.
- On-device: Understand and reason across multimodal data generated on phones, cameras, wearables, and other connected devices.
“Graphon has been an invaluable partner in GS Group’s AI transformation journey, bringing exceptional passion and AI expertise to the table,” said Ally Kim, Vice President at GS, one of the largest South Korean conglomerates. “Their multimodal AI solutions have been instrumental in solving real-world challenges, such as analyzing customer movement in convenience stores and enhancing safety through CCTV analysis at construction sites.”
Jennifer Chayes, Dean of the College of Computing, Data Science, and Society at UC Berkeley, and Christian Borgs, the UC Berkeley computer science professor who coined the term graphon — both Technical Advisors to Graphon — said: “Graphon’s technology automatically discovers relationships across data and treats that structure as a first-class citizen. We’ve known that the future of AI would depend on understanding structure beyond tokens; their continuous multimodal graph representation unlocks a new paradigm of system models.”
For more information, visit graphon.ai.
About Graphon.ai
Graphon.ai is building the pre-model intelligence layer for enterprise AI. By transforming raw multimodal data into persistent relational memory, Graphon enables foundation models to reason over connected systems instead of isolated token sequences. By replacing context windows with structured memory, Graphon improves accuracy, reduces hallucination, and enables LLMs to operate at enterprise scale. Graphon is founded by engineers and researchers from Amazon, Meta, Google, Apple, NVIDIA, and leading research institutions. For more information, visit graphon.ai.
Contacts
Hannah Lulinski
VSC for Graphon
graphon@vsc.co
Featured image: Credit: Graphon AI
