Ashu Garg

Ashu Garg is a general partner at Foundation Capital, investing across enterprise software, AI infrastructure and AI-enabled applications. An early investor in Databricks and Eightfold, he works with technical founders on product-market fit and their development as CEOs. His public writing with Jaya Gupta argues that enterprise AI startups can build lasting value by recording the policies, exceptions and approvals behind organizational decisions.[2][3]

In their December 22, 2025 essay, “AI’s trillion-dollar opportunity: Context graphs”, Gupta and Garg distinguish general rules from decision traces: records of how a rule was applied in a particular case, including evidence, exceptions, approvers and resulting changes. They define a context graph as those traces linked across entities and time, making precedent searchable. It records organizational decisions rather than a model’s chain-of-thought.[3]

Their hypothetical renewal example makes the distinction concrete. An agent proposes a 20% discount despite a policy cap of 10%. It gathers three severe service incidents, a customer’s cancellation escalation and a previous VP-approved exception, then routes the request to Finance. The CRM ultimately records the discount; the decision trace preserves the evidence and approval that justified it. Their proposed starting point keeps humans in the approval loop while agents gather context and record precedent.[3]

Gupta and Garg’s investment argument centers on being in the execution path. An orchestration layer can capture context when an action occurs, whereas, they argue, later warehouse ingestion may preserve the outcome without the decision context. They describe three routes for startups: replacing an entire system of record, replacing a specific workflow while syncing final state to an incumbent, or creating a new system of record for decisions. The trillion-dollar framing is their opportunity thesis.[3]

In his January 30, 2026 follow-up, Garg accepted criticism that true intent is internal and unobservable. Drawing on Arvind Jain of Glean’s distinction, he wrote: “I agree: you capture the how – the policy applied, the evidence consulted, the exception granted, the approver – and infer the why from patterns over time.” This qualifies the original promise to capture why decisions happened: observable process traces provide the basis for inference.[4]

Garg also left implementation questions open: whether context graphs become a standalone category or a feature of broader platforms, where they should live, who may query sensitive traces, how long records should remain, and when old precedents cease to apply. He credited Animesh of PlayerZero with an approach in which schemas emerge through agent use, and Conviva’s Vyas Sekar and Hui Zhang with extending the idea toward stateful reasoning that connects actions to outcomes. He acknowledged that the boundary between encoding organizational knowledge and allowing agents to explore may vary by domain.[4]

Garg’s work on organizational knowledge also included early brainstorming for Viven. In his October 17, 2025 investment announcement, he recalled spending almost every weekend discussing ideas at his home whiteboard or on walks with Ashutosh Garg and Varun Kacholia, Viven’s founding pair. The discussions explored personalized employee AIs that could answer questions on their behalf and reduce coordination costs. He said those conversations grew into Viven, which Foundation Capital supported from its earliest stage.[6]

Ashutosh Garg, who also goes by Ashu, is a different person: the Eightfold and Viven founder. The Foundation Capital investor had previously backed Eightfold. His Viven announcement describes a platform intended to ground employee digital twins in emails, documents, code, messages and meetings, with relationship-sensitive responses and controls over information sharing. His contribution was as an investor and brainstorming participant alongside the founders.[6]

Garg’s enterprise AI investing predates the context-graph essays. Looking back in “Reinventing Enterprise Software,” published November 29, 2017, he recalled using machine learning for behavioral targeting while running Microsoft’s online-advertising business. Soon after joining Foundation Capital in 2008, he joined Conviva’s board. He described that exposure as giving him an early view of Spark’s potential and informing his decision to become one of Databricks’ first investors when it formed to commercialize Spark.[5][2]

His investment in Aggregate Knowledge supplied another example: using longitudinal advertising data for multi-touch attribution rather than assigning credit only to the last interaction. In his account, both Aggregate Knowledge and Conviva needed custom data infrastructure. Garg said these investments helped shape his thesis that machine learning could reinvent enterprise software as compute, storage and tooling became more accessible. In 2017, he identified DevOps, recruiting, sales and finance as areas where algorithms could automate repetitive, data-intensive work.[5]

Before Foundation Capital, Garg led Microsoft’s online-advertising business and field marketing for its software businesses, and worked at McKinsey on technology companies’ go-to-market efforts. Earlier roles included founding TringTring.com, establishing Unilever’s Nepal operations, and leading marketing and pre-sales teams at Cadence Design Systems. Foundation Capital’s biography gives a concrete example of his commercial assistance: in 2010, he helped TubeMogul’s small team reach brands moving television advertising online as the company expanded from analytics into media buying. He also hosts B2B a CEO, a podcast for technical founders whose guests have included Eric Yuan, Ali Ghodsi, Jennifer Tejada, Aaron Levie and Frank Slootman.[2]