Context graphs are gaining traction as a significant concept in AI, according to Ashu Garg and Jaya’s recent analysis. They argue that these graphs function as an organization's institutional memory, capturing how decisions are made rather than simply documenting outcomes. Key industry figures have echoed this sentiment. For instance, Dharmesh Shah emphasizes that context graphs serve as a decision-making record, while Aaron Levie notes that as models become ubiquitous, the real differentiator will be the organizational knowledge fed into them. The authors highlight that traditional enterprise software excels at recording outcomes but fails to capture the reasoning behind decisions, which often resides in informal channels like Slack or personal memory. They introduce the term "decision traces" to describe this missing context, which can accumulate into a searchable context graph over time. This repository of decision-making processes can provide companies with a proprietary advantage, as the knowledge is difficult to replicate. Garg and Jaya believe that startups focusing on capturing these decision traces will hold an edge over established companies that struggle to retrofit their systems. Concerns about the feasibility of capturing the reasoning behind decisions are acknowledged. Skeptics argue that true intent is often internal and unobservable. However, the authors assert that tracking the sequence of actions—the "how"—can effectively approximate the "why" over time. This approach enables agents to operate within established organizational rules while still allowing some flexibility. The piece concludes by suggesting that the concept of context graphs may evolve, possibly becoming a core feature of broader platforms, but the opportunity for startups to create valuable assets remains.