Evolution is the problem. AI is the answer.

A buying committee now has 9 to 13 people. Your best rep can hold the details of maybe two deals that complex in their head at a time. So they do something completely rational, and costly: they retreat to the relationships and products they already know.

A few weeks ago I wrote about Dunbar's Number — the idea, grounded in evolutionary biology, that the human brain can actively manage only about 150 relationships before accuracy starts to break down. The response from sales leaders was immediate, and it almost always took the same form: "That explains something I've been watching for years and couldn't quite name."

So let me name it. Here is what Dunbar's ceiling actually looks like on your sales team — not in theory, but in the daily behavior of your best people. An evolutionary artifact that is costing you deals today.

A modern enterprise deal is not one relationship. It's a web. Gartner now puts the average B2B buying committee at 9 to 11 people. Forrester puts it at 13 internal stakeholders plus another handful of external advisors. Each of those people has motivations, agendas, alliances, a history with your company, and a relationship to every other person in the room. That's not 13 relationships. Counting the connections between them, it's closer to a hundred.

Do the arithmetic and it stops being abstract. The relationship load of a single healthy pipeline doesn't approach Dunbar's ceiling — it blows through it, several times over. The human brain your top performer is running on was optimized for a tribe of 150 people who all knew each other. It was not built for a CRO's pipeline. The hardware hasn't changed. The job has.

It's worth asking why the buying committee exploded in the first place. It's a story about how authority inside organizations has fundamentally changed.

A generation ago, leadership meant something specific: an executive held both the power and the authority to decide. For better or worse, that model is gone. Authority has been distributed, diffused, and defended. Decisions that once belonged to a single leader now run through consensus, committee, and a sprawl of stakeholders who each hold a veto but none of whom hold the pen.

For the seller, the consequence is brutal and underappreciated. The number of relationships required to close a deal didn't grow incrementally — it grew exponentially, as organizations traded individual authority for collective consensus. We engineered the buying committee into existence, and the buying committee is precisely what overwhelms the human cognitive ceiling.

When a system exceeds its capacity, it doesn't fail randomly. For human beings under cognitive overload, that direction is well documented: we retreat to what is familiar and low-effort. We conserve.

For a salesperson, "familiar and low-effort" has a specific meaning. It means the stakeholder who always takes the call. The champion they already trust. The two or three products they know cold. Faced with a buying committee of thirteen strangers and a catalog of forty products, the overloaded rep doesn't expand. They contract.

This is the behavior every sales leader recognizes and most misdiagnose. The rep who won't multi-thread the deal. The rep who pitches the same three products on every opportunity. The rep who "has great relationships" but somehow never widens them. We label this a coaching problem, a motivation problem, a discipline problem. It is none of those. It is a capacity problem. The rep retreated to what they knew because what they knew was the only part of the deal their brain could still hold.

This is why the usual fixes don't move the number. More training doesn't add working memory. A better CRM doesn't reason over relationships — it stores fields and waits to be searched, which adds retrieval work to a brain that's already maxed. An AI copilot that summarizes a call doesn't tell the rep which of the thirteen stakeholders just became the risk. Every one of those tools assumes the bottleneck is information access. It isn't. The information was always there.

The only thing that actually moves the ceiling is removing it. And there is exactly one kind of mind that was never bound by it in the first place.

A knowledge graph doesn't get tired, doesn't suffer recency bias, and doesn't retreat to what's comfortable when the load gets heavy. It holds every stakeholder in every deal at full resolution — who influences whom, who's gone quiet, who just changed roles, which relationships are warming and which are cooling — across the entire pipeline, simultaneously.

AI doesn't replace the seller. People close deals — through trust and judgment no machine can manufacture. What the graph does is end the retreat. It hands the rep the full map of the deal so they never have to fall back to the one stakeholder they remember or the three products they know. It tells them who else matters, what changed overnight, and what to do next — pushed to them, not waiting to be searched for.

Pull the threads of this piece together, and each failure of the human cognitive ceiling has a direct architectural answer.

None of this replaces the seller. People still close deals through trust and judgment no machine can manufacture. What changes is that the part of selling that is pure scale — remembering, watching, matching, connecting — is finally handled by something built for scale. The human does the human part. The machine does the rest.

The next time you see a rep selling to the same people the same way, don't reach for a coaching plan. Reach for the realization that they've hit a wall every human brain hits. The question isn't how to push them through it. It's whether you're going to keep asking them to.

Your reps have a Dunbar Number. Your competition's AI doesn't. Neither does ours.