Tribe
Why the company you build for an AI-driven world looks nothing like the one you started.
Hey,
I want to start with a confession that took me ten years to be ready to write.
We built the best Account or Client Managers in the world.
I am not exaggerating. If you put one of our AMs in a room with the head of growth at a Series B SaaS company, our person will outthink them on ICP, outexecute them on sequencing, outcoach their SDRs, and outstrategize their next quarter.
We have spent a decade sculpting these humans. They are part consultant, part operator, part coach, part closer. They are the reason Belkins exists at the scale it does. They are the reason our retention numbers embarrass our category.
And they are exactly the reason we hit a wall.
The AM we built is a unicorn. Not one in a thousand. Maybe one in ten thousand.
We grow them internally over two to three years, because the talent market does not produce them at hireable compensation. Each one takes a slice of my attention, our senior team’s time, and the company's institutional memory to fully form.
A unicorn does not scale.
You cannot hire fifty of them next year. You cannot find them on LinkedIn. You cannot 10x the company by 10x-ing the headcount of a role that takes years of internal sculpting to produce.
We tried. For a long time. And every quarter the math got tighter.
Last quarter we had nine clients on what our team calls the Predicted Churn list. Nine accounts quietly drifting. When I went back through the notes, each one had the same root cause.
Their AM had run out of hours in the day.
Not bad AMs. Some of our best. Each running point on a dozen relationships, debugging deliverability, coaching SDRs, sitting in QBRs, trying to find an hour to actually think. The math had stopped working months earlier. We just hadn’t admitted it yet.
That is the moment I stopped pretending the AM role at Belkins was hard, and started admitting it was impossible. Not because the people are not extraordinary. Because the role itself was designed for a smaller company that no longer exists.
This edition is about what we are replacing it with.
A new structural unit. A new social unit. A new operating system underneath both. And, most importantly, a new word for what the company is becoming.
The word is Tribe.
The unicorn problem
What the AM job description at Belkins actually requires, if you write it honestly.
Strategic consulting in the morning, the kind of conversation a CMO has with a partner at McKinsey.
Delivery operations at lunch, the kind of triage a head of customer success runs at a SaaS company.
SDR coaching in the afternoon, the kind of work a sales manager does.
Senior client relationships by six in the evening, the kind of judgement a VP of sales brings to a renewal conversation.
Each of those is its own discipline. Each requires its own headspace, its own training, its own success metrics. The talent market does not produce humans who can do all four at hireable comp.
So we made our own. For a decade. And we got incredibly good at it.
The problem is that being incredibly good at producing unicorns is not the same as being able to scale a company. It is the opposite. The better we got at sculpting the role, the more dependent we became on a production process that simply does not run faster than two or three years per human.
The math is brutal once you write it down:
A healthy AM load is twelve to fifteen clients. We are adding roughly twenty new clients every month. Even at full capacity, we run out of room in ninety days.
Adding more AMs does not fix this. Even if we could find them, the role itself caps how many clients each one can serve.
We were not facing a hiring problem. We were facing a design problem.
Once you see it as a design problem, the answer becomes obvious. You stop hiring for the union of four jobs. You split the role into specialised lanes. You give each lane an AI counterpart. You wrap the whole thing in a structural unit small enough to keep its rhythm but large enough to compound.
You build a pod.
From Circles to Tribes
Some of you read about Circles; if not, please do.
I wrote about Folderly. Circles is a Holacracy-inspired model where roles replace titles, teams self-govern, and authority flows from accountability rather than from hierarchy.
Folderly still runs on Circles today, and the team loves it.
It works there because Folderly is a software company, where the unit of work is a problem, not a client. Circles handle problems beautifully.
Belkins is different. The unit of work at Belkins is a client relationship that lasts two to three years, demands daily delivery, and requires a stable cast of people who know each other.
Circles alone do not give you that. You need a structural unit underneath the cultural one.
That is where the pod comes in.
A pod is six to ten SDRs, one Account Strategist, one Delivery Lead, and one AI Account Assistant. It serves eighteen to twenty active clients. It has its own rhythm, its own metrics, its own tiny culture.
Five pods serve sixty to ninety clients on the same payroll as the five AMs we have today.
Now stack the pods. About eight of them, plus a Tribe Lead, makes a tribe. Roughly 150 active clients per tribe.
The number is not arbitrary.
Why “Tribe”
I tested the other words first. Squad. Cluster. Band. Group. Guild. None of them carried the right weight.
A squad is military. Replaceable parts moving on command. The unit dissolves when the mission ends.
A cluster is what data scientists call points on a graph. It has no identity, no scars, no shared memory.
A band is small but cult-like. Bands break up when one member leaves.
A guild is a vertical association of one craft. Guilds standardize quality but do not produce belonging.
A tribe is something older and more durable than all of them.
Sebastian Junger writes about this beautifully. Humans evolved in groups of 30 to 50, within bands of about 150. The number 150 is roughly Dunbar’s cognitive ceiling on stable relationships, the largest group in which you can know everyone’s name and remember what they care about.
Tribes share three things that none of the alternatives do.
They share identity. A tribe member knows which tribe they belong to and can tell you why. The tribe has a story about itself.
They share scars. A tribe remembers what it has survived together, and that memory becomes the bond. New members earn their way in by sharing the next scar.
They have the right to expel free riders. The tribe is small enough that contribution is visible. You cannot hide. Reciprocity is the contract.
That is the unit I want at Belkins. Eight pods, one Tribe Lead, 150 clients, all of them inside an organizational unit small enough to feel like one team. Big enough to absorb a hundred million in revenue. Stable enough to survive a founder who is absent on a Wednesday.
The pod is the family. The tribe is the band. The platform is the civilization. We are building all three at once.
The marketplace reframe
Once you have the unit cell, you start to see the company differently.
Belkins is not an agency. Belkins is a marketplace.
SDRs supply attention, execution, and time. Clients demand qualified meetings. The job of the company is to match supply to demand at the highest possible quality, every day, at scale.
Historically, the AM was the matching layer. A human bottleneck carrying ICP knowledge, sequence playbooks, and relationship context inside their head.
That is the part of the system AI replaces. Not the relationship. Not the strategy. The matching.
We are building the matching layer as a platform, not a person. The platform is called Belkins Home. Every SDR queue, every campaign, every SQL handoff, every AI augmentation lives inside one system. It is, structurally, the rail that routes supply to demand.
Once the rail exists, the AM role decomposes naturally:
