Graphon Labs, a San Francisco startup that builds memory infrastructure for AI agents, raised $35m in a Series A round led by RRE Ventures, it announced on 10 November 2026.
HP, Energy Impact Partners, Hitachi Ventures and Gaia VC also took part, according to The Next Web and Unite.AI. The company did not disclose a valuation.
The round came six months after Graphon announced its seed funding. The company put its total funding at $43m; The Next Web and Unite.AI gave the figure as $43.3m.
“Graphon Labs is ushering in software as memory, where what you keep is understood, not just stored.”Arbaaz Khan, founder and CEO, Graphon Labs
What Graphon Labs does
Graphon sells software that gives AI systems a lasting memory of the material they are fed. It takes in documents, spreadsheets, images, audio and video, builds a graph of how the pieces relate, and keeps track of where each item came from and when it changed. The name refers to graphons, a mathematical tool for describing very large graphs, which the company uses in compressed form to steer retrieval.
Developers work with the system through an interface modelled on Amazon S3 storage, with buckets, objects and queries. Users can run keyword, semantic and image searches, or put questions to an agent that follows links between sources. The company says it holds SOC 2 Type II certification.
It offers three products, according to Unite.AI: Graphon Enterprise, which can run in Graphon's cloud, in a private cloud or on a customer's own servers; a developer platform; and on-device memory for phones, laptops, cameras and wearables. The Next Web reported that the San Francisco 49ers and Hitachi are customers, and cited Business Insider as saying HP is one too. The 49ers are exploring uses such as finding sponsorship deals, while Hitachi Rail described an evaluation in an industrial setting as promising.
The company also published its own benchmark results: 93.9% on LoCoMo, a test of memory over long conversations, while using 58% fewer tokens than loading the full conversation, and 81.7% on ATM-Bench, a University of Cambridge benchmark spanning video, audio, images and text. These are company-reported figures.
Who is behind it
Graphon was founded in July 2025, according to Unite.AI, citing its company profile. Its chief executive, Arbaaz Khan, and chief technology officer, Clark Zhang, built graph-based memory systems for robots during their doctoral studies, The Next Web reported. Deepak Mishra is chief operating officer. The company has about 20 staff.
Raju Rishi, a general partner at RRE, said in comments reported by Unite.AI that models can reason and agents can act, but that neither improves without the ability to build on earlier knowledge.
Khan described the company's aim as making software that understands what it keeps rather than simply storing it.
What the money is for
Most of the money will go on hiring engineers and researchers, with some spent on sales staff, according to The Next Web.
Graphon plans to serve developers and businesses first, then extend the system to cloud and on-premises deployments and to consumer devices. In a blog post announcing the round, Khan wrote that he wants to be able to ask questions, in his own words, of the photos, videos and conversations people keep.
The funding so far
Graphon came out of stealth in May 2026 with an $8.3m seed round, announced on 14 May and led by Arvind Gupta of Novera Ventures, according to Unite.AI. At that stage it named the South Korean group GS as an early enterprise customer.
With the Series A, the company has raised about $43m in roughly six months.
Company website: graphonlabs.ai
