Regie.ai

Context Graphs Are Real — Sales Is One Place They Become Concrete The idea of context graphs, thanks to our very own Jaya Gupta and Ashu Garg has rightly captured a lot of attention. Foundation Capital’s “Context Graphs: AI’s Trillion-Dollar Opportunity” lays out a compelling, aspirational vision: AI systems that move beyond stateless systems and instead operate with durable, evolving context over time. It’s a powerful north star. At the same time, as Dharmesh Shah pointed out — the original framing from our investors and board members, Foundation Capital is intentionally abstract and aspirational. It can make the idea feel hard to operationalize in the real world as Dharmesh noted. At least on first blush. Sales is one of the most concrete, observable, and operational domains where context graphs can be built today even with Dharmesh's pointed limitations. Why Sales Is a High-Signal Domain for Context Graphs Sales has three properties that make it a strong proving ground: 1. State changes constantly — accounts evolve, stakeholders change, priorities shift. 2. Context is fragmented — spread across CRM, email, calls, Slack, sales rooms, and product usage. 3. But outcomes are measurable — pipeline progression, deal velocity, retention, expansion. Not an exercise in context graph philosophy. The problem, as been pointed out is that CRMs don't preserve state. CRMs were never designed to be systems of memory. They store fields, not history. They store snapshots, not trajectories. For example as it relates to demand gen: When was an account last enriched? When did an email address change? Was that change driven by a rep, automation, or data vendor? Was the new information ever acted upon? In Salesforce, even tracking when a phone number or email field was updated requires explicit admin configuration — and even then, the historical view is partial. The key difference though is that ironically all the sales channels and their associated systems already maintain historical state. This is where I think it makes things easier than Dharmesh might point out. The systems sales teams rely on every day already preserve rich state. Slack and email captures communication over time including people and personas. Gong preserves conversational history, objections, momentum. Digital sales rooms tools track buyer interaction over time. Data warehouses store historical product usage. These systems don’t just log final state — they retain contextual memory. Context graph in sales starts here. This is the first of a series of posts! I've been on a LinkedIn hiatus because we've actually been building on Context Graphs thanks to Foundation Capital and Ashutosh Garg and several of our select customers who have done 100+ hours of hands-on work on the topic! And we've been heads down here on product development over the last 6 months. I'll try to catch you up on this series!

Most sales “tribal knowledge” never becomes reusable. Your best sellers don’t just know what to do — they know why. They recognize patterns, remember exceptions, know who to pull in, and understand what worked last time in a similar situation. That context is the real advantage… and it usually lives in people’s heads, Slack threads, and one-off decisions. This Foundation Capital piece by  Jaya Gupta and Ashu Garg helped crystallize the missing layer: once you have decision records, the “why” becomes first-class data. Over time, these records naturally form a context graph — the entities the business already cares about (accounts, renewals, tickets, incidents, policies, approvers, agent runs) connected by decision events (the moments that matter) and “why” links. That’s what makes agents powerful. Agents aren’t valuable because they can “write an email.” They’re valuable when they can tap into — and contribute to — the collective knowledge of your team: -pulling the right context across systems -proposing the next best action based on precedent -escalating the right moments to humans -and recording the rationale so the org gets smarter over time When “why” is stored and connected, companies can finally audit and debug autonomy, and turn exceptions into precedent instead of re-learning the same edge case in Slack every quarter. That’s the future of sales: humans + agents, with context and judgment compounding. Link in comments.