Jaya Gupta is a partner at Foundation Capital, where she invests in enterprise software and AI applications. Her essays with Ashu Garg examine where durable business value sits when model capabilities and products change quickly.[1]
Start with AI’s trillion-dollar opportunity: Context graphs →
X · jayagup10 Career
Gupta first joined Foundation Capital as chief of staff, working on portfolio operations and strategy. Before that she worked in McKinsey’s technology practice on software diligence and startup go-to-market plans. Foundation’s biography says she built CardioBuddy while studying at Georgia Tech and later worked at startups that were acquired. Her current investing work spans the enterprise stack, including the infrastructure underneath AI applications.[1]
Ideas
A record of why a company acted
In the context-graphs essay, Gupta and Garg distinguish a company’s stored business objects from its decision history. A customer database can show an approved price without preserving the exception, precedent and approval that produced it. They argue that agents placed inside workflows can capture those traces as decisions happen, linking them across people, cases and time. Their investment thesis is that this accumulated context could become a defensible asset for application companies. It is a proposed source of business advantage, rather than a benchmark proving that every agent learns reliably.[2]
Competing with the model supplier
Her Service-as-Software essay addresses the risk that an AI application’s model supplier becomes its competitor. The authors favor systems that improve through domain-specific interactions and workflow knowledge. This shifts the question from whether a startup has a clever prompt to whether its work produces information and operational learning that a general model provider cannot easily reproduce.[3]
Tools, writing & talks
Writing, research & talks
Start here
AI’s trillion-dollar opportunity: Context graphsfoundationcapital.comAn argument for preserving decision traces, such as exceptions and approvals, as a new layer of enterprise software.
Read the context-graphs essay first. It uses familiar business decisions to introduce the thesis without requiring machine-learning background; ask which decisions your own systems currently fail to preserve.
Co-author with Ashu Garg[2]An investor essay argues agent workloads are shifting from inference to execution, driving CPU, storage and sandbox demand. It claims power moves to the companies assembling agents and routing between models, runtimes and suppliers, with CoreWeave's financing strain as the warning.
Catalog workOriginal source- When model providers eat everythingfoundationcapital.com
A strategy essay about building AI applications that retain an advantage as model suppliers enter application markets.
Co-author with Ashu Garg[3]