Jaya Gupta & Ashu Garg — Foundation Capital - AI’s trillion-dollar opportunity: Context graphs - December 2025
FreeFreeFree tier
About Jaya Gupta & Ashu Garg — Foundation Capital - AI’s trillion-dollar opportunity: Context graphs - December 2025
Foundation Capital partners Jaya Gupta and Ashu Garg argue that the next trillion-dollar enterprise opportunity lies not in improving systems of record (Salesforce, Workday, SAP) or simply building better AI agents, but in creating a new category: systems of record for decisions. They introduce the concept of a 'context graph'—a queryable, living record of decision traces (exceptions, overrides, precedents, cross-system context) that currently exist in Slack threads, deal desks, and people's heads. Context graphs would provide agents with the organizational memory needed to handle real-world ambiguity, moving beyond rules and governance to capture how rules were actually applied, where exceptions were granted, and who approved what. This perspective was published on the Foundation Capital blog in December 2025.
Key Features
Argues that agents need 'decision traces' (exceptions, overrides, precedents) not just rules and governance
Introduces the concept of 'context graphs' as a living record of decision traces stitched across entities and time
Explains that context graphs make precedent searchable and become the real source of truth for autonomy
Contrasts rules (what should happen) with decision traces (what actually happened in specific cases)
Posits that systems of record for decisions could become the next trillion-dollar platforms
Pros & Cons
Pros
- Offers a clear, well-argued thesis about the next major enterprise category
- Provides specific, actionable examples of decision traces (e.g., definition exceptions, policy versioning, approval chains)
- Acknowledges and builds upon existing analysis (Jamin Ball's 'Long Live Systems of Record') creating a nuanced discussion
- Accessible to both technical and business audiences without being oversimplified
Cons
- The concept of 'context graphs' is presented as a theoretical framework, not a concrete product or implementation
- Does not address potential challenges like data privacy, security, or storage costs of persisting all decision traces
- Focuses entirely on the opportunity side without discussing barriers to adoption or incumbent responses
Best For
Understanding the structural limitations of current systems of record for AI agent workflowsIdentifying startup opportunities in building context graph infrastructure for enterprise automationInforming architectural decisions for agentic systems that require organizational memory and exception handlingProviding a framework for investors evaluating AI infrastructure startups
FAQ
What is a context graph according to this article?
A context graph is an accumulated structure formed by decision traces—a living record of how decisions were made across systems, including exceptions, overrides, precedents, and cross-system context. It is not a model's chain-of-thought but a queryable record of decision traces stitched across entities and time, making precedent searchable.
Why do agents need decision traces instead of just rules?
Rules tell an agent what should happen in general, but decision traces capture what happened in this specific case—which definition was used, under which policy version, with what exceptions or approvals. Agents need access to these traces to handle the ambiguity that humans resolve with judgment and organizational memory.
How does this relate to systems of record like Salesforce or Workday?
The article argues that agents do not kill existing systems of record, but raise the bar for what a good one looks like. However, it goes further by proposing that entirely new systems of record will emerge—systems of record for decisions rather than just objects—and these could become the next trillion-dollar platforms.
Who wrote this article?
The article was written by Jaya Gupta and Ashu Garg, partners at Foundation Capital, a venture capital firm, and published on their blog in December 2025.