Relationships don't just matter to humans. They're the entire architecture of intelligence.

Executive Summary

A growing body of research in evolutionary biology, cognitive science, and organizational behavior converges on a single insight with profound implications for enterprise sales technology: human cognitive architecture imposes a hard limit on the number of complex, interconnected relationships any individual can actively manage.

This limit — known as Dunbar's Number, approximately 150 stable social relationships — was adaptive for the small-group environments in which the human neocortex evolved. It is catastrophically mismatched to the demands of modern enterprise sales, where a single complex deal may involve 9 to 13 stakeholders connected through influence networks, organizational hierarchies, competitive dynamics, and temporal shifts that change weekly.

The persistent performance gap between top-performing and average sales reps — the "80/20 problem" — is fundamentally a cognitive architecture problem, not a training or tooling problem. Top performers succeed because they can hold more relationship complexity at higher resolution than their peers. This capability is scarce because it is bounded by biology.

Four generations of CRM and sales technology — systems of record, conversation intelligence, revenue intelligence, and AI copilots — have failed to address this gap because they process data about relationships rather than modeling the relationships themselves.

SAIQ's Generation 5 architecture solves this through a persistent knowledge graph that models relationships as first-class objects across four dimensions: people, knowledge, time, and context.

In 1993, British anthropologist Robin Dunbar and American anthropologist Leslie Aiello published a landmark paper in Current Anthropology titled "Neocortex Size, Group Size, and the Evolution of Language."¹ By studying the correlation between neocortex volume and average social group size across 38 primate species, they proposed that the human neocortex imposes a cognitive limit of approximately 150 stable social relationships — relationships in which an individual knows who each person is, how they are connected to others, and what behavior to expect from them.

The number 150 is not arbitrary. Dunbar and subsequent researchers found it reflected in a striking range of human organizational structures: Neolithic farming villages, Roman army units (the century), Hutterite communities, average personal social networks, and modern military companies. Dunbar also identified a fractal structure of nested layers — about 5 intimate relationships, 15 close friends, 50 in the broader sympathy group, and 150 in the active social network.³

In 2021, researchers at Stockholm University led by Patrik Lindenfors rigorously retested Dunbar's hypothesis. Their analysis, published in Biology Letters, found that confidence intervals ranged from 4 to 520, leading them to conclude that "specifying any one number is futile."⁴ They did not, however, dispute the existence of a cognitive ceiling — only the precision with which it could be specified.

This ceiling is an evolutionary inheritance. It was optimized through millions of years of primate evolution for small-band environments where managing alliances, threats, and cooperative obligations conferred survival advantage. The social brain hypothesis has been validated across primates, ungulates, carnivores, bats, cetaceans, and birds.⁵

The modern enterprise deal imposes cognitive demands that far exceed the environment for which the human neocortex was optimized. According to Gartner's 2025 research, the average B2B buying committee now includes 9 to 11 stakeholders, up from 5 to 7 in 2017⁶ — a near-doubling in under a decade. Forrester's State of Business Buying 2024 report places the figure even higher at 13 internal stakeholders plus 9 external participants per average purchase decision.⁷

Each stakeholder exists within a web of organizational relationships — reporting lines, influence networks, political alliances, historical interactions — that the selling team must understand and navigate. A single enterprise deal is a relationship graph of 50 to 100+ nodes when you account for the interconnections between stakeholders, competitors, internal champions, product requirements, and historical deal context.

The conventional interpretation is that most reps need better training, better tools, or better management. The data tells a different story.

Only 28% of sales reps hit annual quota in Salesforce's 2025 State of Sales report — the lowest figure in six years.⁸ RepVue's Q2 2025 data shows 57.31% missing quota.⁹ When a top performer leaves, the relationship intelligence leaves with them. Enterprise AEs take 7 to 9 months to reach baseline productivity, with top-performer status requiring 15 to 18 months¹⁰ — not because the product is hard to learn, but because the relationship graphs that defined the predecessor's effectiveness must be rebuilt from scratch in a new brain.

Four generations of CRM and sales technology have attempted to address the performance gap. Each has added capability. None has addressed the underlying cognitive constraint. The common failure across all four is architectural: they process data about relationships without modeling the relationships themselves.

A contact record that says "Champion" is a label, not a relationship. A relationship has directionality (who influences whom), temporality (how engagement has changed over time), weight (active champion versus going silent), context (what do they care about, what deal history exists), and interconnection (how does this person's position relate to every other position in the deal). No generation prior to the fifth has built the architecture to represent and reason over these dimensions.

SAIQ's Generation 5 architecture was designed from the ground up to answer a specific question: what would it take to represent every relationship that matters in a deal — and reason over all of them simultaneously?

The answer is a persistent knowledge graph in which relationships are first-class objects, not fields in a database. The graph models four categories of relationships, each of which reflects a dimension of cognition that human brains perform intuitively at small scale but cannot sustain across the full scope of an enterprise pipeline.

People relationships

In the graph, a person is not a record. It is a node connected to dozens of other nodes through typed, weighted, temporal edges. Who do they influence? Who influences them? What is their engagement trajectory — accelerating or decelerating? How does their behavior in this deal compare to their behavior in previous deals? Are they connected to anyone at the competitor? Have they changed roles, and if so, what does that mean for their authority in this purchase? Each question is answered by the relationships extending from the node, not by the node itself.

Knowledge relationships

Deals do not happen in isolation. Every opportunity inherits knowledge from every previous opportunity at that account, in that industry, against that competitor, using that product configuration. SAIQ's commercial ontology encodes knowledge relationships that would otherwise exist only as institutional memory — which product capabilities map to which buyer pain points, which pricing structures have succeeded with this persona, which competitive positioning has worked against this incumbent.

A top-performing rep carries a partial version of this in their head, built from years of pattern matching. SAIQ's graph carries the complete version — across every deal, every rep, every outcome — and makes it available to every team member on day one.

Temporal relationships

This is the dimension nearly every AI tool ignores, and it is arguably the most important. Every relationship in a deal exists in time. The champion who was engaged three months ago may be disengaged now. The competitive threat that did not exist in Q1 may be dominant in Q3.

Confabulation causes us to fill gaps in memory with plausible but inaccurate narratives. Conflation causes us to merge details from different deals because the situations felt similar. Both degrade relationship accuracy over time. SAIQ's graph maintains a versioned, temporal representation of every relationship. When delta reasoning is applied to a new signal, the system compares it against the full temporal pattern of every related relationship to determine what changed, what it means, and what action is warranted.

Contextual relationships

Context is the web of relationships that gives any individual fact its meaning. The fact "CFO attended the meeting" is meaningless without context. The graph connects that fact to the CFO's attendance pattern (they usually do not attend), the stage of the deal (late-stage), the champion's recent silence, the company's cost-reduction mandate, and the competitor's price cut three days ago.

Without the graph, the CRM logs "CFO attended." With the graph, SAIQ generates a contextual assessment: the CFO's unprecedented late-stage attendance, combined with champion disengagement and a cost mandate, suggests executive budget scrutiny — possibly triggered by competitive pricing pressure. Recommended action: executive sponsor alignment within 48 hours, reframing value around cost avoidance.

That is not a summary. It is contextual reasoning across a relationship graph. It is the kind of reasoning top-performing reps do intuitively — on the deals where they have sufficient cognitive capacity to hold all the relationships in their head.

Over millions of years, the primate neocortex evolved to solve a specific problem: managing social relationships in increasingly large groups. As Dunbar and Shultz documented across extensive comparative research, there is "an extremely robust statistical relationship between the typical size of a species' social group and the size of its neocortex."⁵ The neocortex did not evolve to process data. It evolved to process relationships.

The neocortex did not replace the limbic system, the brainstem, or the sensory cortex. It integrated their signals into a higher-order representation of the social world. Generation 5 does not replace your CRM, your conversation intelligence, your forecasting tools, or your AI copilot. It integrates their signals into a higher-order representation of the deal — a representation structured around relationships rather than records.

The human brain manages approximately 150 relationships by instinct — the product of millions of years of evolutionary optimization, running on a biological computer that consumes approximately 20 watts of metabolic power¹¹ and requires eight hours of nightly restoration. SAIQ's graph manages 150,000 relationships by architecture. Or 1.5 million. The number is irrelevant.

But the goal is not to replace the human. Graphs do not close deals. People close deals. The relationships between a seller and a buyer are human relationships, governed by trust, empathy, credibility, and judgment that no computational architecture can replicate.

The goal is to solve the 80/20 problem. Top performers have both the skill to know what to do next on a deal and the will to actually do it. They process relationship complexity at a level their peers cannot match. SAIQ's graph captures that processing — the relationship maps, the pattern recognition, the contextual reasoning, the temporal awareness — and delivers it to every representative, on every deal, every day.

A new hire does not start at zero. They inherit the full relationship graph of every deal they touch. Not because they memorized CRM notes. Because the architecture remembers what no human brain can hold.

Every generation of sales technology has attempted to solve the performance gap between top performers and everyone else. Each has added more data, more visibility, more tools. The gap persists because the gap was never about data. It was about relationships.

Top performers do not outperform because they have access to better information. They outperform because their brains can hold more relationship complexity at higher resolution than their peers. That is an evolved cognitive capability. It is real. And it is scarce.

The question for every revenue leader is whether to accept that scarcity as permanent — or build an architecture that makes the capability available to every member of the team.

Dunbar's Number was the ceiling. The graph is the breakthrough.