In a rapidly evolving enterprise AI landscape, venture capitalists are coalescing around a new architectural thesis: context graphs. Coined in a December 2025 essay by Foundation Capital’s Jaya Gupta and Ashu Garg, the concept frames context graphs as systems that record not just what happened in business workflows, but why—capturing decision traces, exceptions, approvals, and precedents across time and systems. This, they argue, is the missing layer that could underpin the next generation of enterprise platforms—and unlock a trillion‑dollar opportunity. (forbes.com)
Unlike traditional systems of record that store only current state, context graphs stitch together the reasoning behind decisions—who approved a discount, under what policy, and based on which precedent—creating a queryable, structured memory of organizational judgment. Gupta and Garg position this as a leap beyond the $200 billion SaaS market, targeting the $4.6 trillion enterprises spend annually on human judgment and services. (forbes.com)
The timing is propitious. AI agents are increasingly embedded in enterprise workflows, generating decision checkpoints—actions proposed by AI and approved or modified by humans—that naturally produce structured decision traces. VCs see startups that sit in this execution path as having a structural advantage over incumbents, whose architectures are built for current‑state storage and lack decision‑state capture. (forbes.com)
The idea has gained rapid traction. By early 2026, the context graph thesis had sparked widespread industry debate. Atlan’s CEO Prukalpa Sankar described context graphs as the trillion‑dollar opportunity enabling AI agents to understand past decisions and business logic, while arguing that universal context platforms—those integrating across heterogeneous enterprise systems—will win over vertical agents. (atlan.com)
Analyst firms are taking note. Gartner predicts that by 2028, the majority of enterprise AI agent systems will be built on context graph foundations, as organizations seek to close AI’s institutional memory gap by embedding decision logic, workflows, and tribal knowledge into their infrastructure. (atlan.com)
Meanwhile, the ecosystem is coalescing. A February 2026 overview of the context graph landscape highlights key players: Glean, Atlan, DataHub, Squirro, and startups like TrustGraph, Graphlit, Zep/Graphiti, Wayfound, and Cognition AI. These companies are racing to become the system of record for decisions, with some already shipping context‑graph capabilities or positioning themselves as integrators across enterprise systems. (contextgraph.tech)
Still, challenges remain. The category lacks a clear market leader, and foundational capabilities—such as predictive reasoning (“if we structure the deal this way, what is likely to happen?”) and permissioned inference—are underdeveloped. Incumbents like Salesforce and Workday are building agent layers atop legacy architectures, but may be structurally constrained. The question for investors is whether context graphs are a product problem or an architectural one—and whether startups can build from the write path out. (forbes.com)
In sum, context graphs have emerged as a compelling VC thesis: a structured, decision‑centric layer that could transform AI agents from tools into institutional memory. If realized, it may well become the trillion‑dollar foundation of enterprise AI.