The Cell
Roles do not work alone. Five human owners form the smallest unit that can carry full accountability.
By Filip Blagojević7 min readOriginally on Substack
Roles don’t work alone.
In my Jobs of the Future series, I mapped five roles that remain fundamentally human as AI absorbs execution-level work. Each article went deep on a single function. The response was strong. People recognized themselves. They saw where they fit.
But here’s what I didn’t address: none of those roles works in isolation. A brilliant strategist without someone to execute is a philosopher. A great executor without strategic clarity is building the wrong thing fast. A storyteller without substance to tell stories about is a bullshit artist.
The roles are organs. Individually, they’re interesting. Together, they form an organism.
I call that organism the Cell.
What the Cell actually is
The Cell is the smallest viable unit of human organization that can function with full accountability in the AI age. It’s not a team. It’s not a department. It’s a self-contained organism where every critical human function has a dedicated owner with real authority.
Five core functions that must be covered. One or two flexible seats for mission-specific expertise when the Cell operates at full size. Five to seven people maximum in its ideal form - though it can run with fewer.
Each person holds a defined domain with final authority over that domain. No committees. No shared ownership. No “we’ll figure out who decides later.” Every function has a name attached to it.
The Cell scales not by adding humans, but by adding AI leverage underneath each function. One Execution Orchestrator with the right AI toolkit can deliver what a traditional team of 10 produced. The humans stay small. The output stays massive.
This is the core idea. A fixed nucleus of high-level generalists, each owning a critical control plane, leveraging AI to multiply their capacity by orders of magnitude.
The five core functions
These are non-negotiable. Every Cell must have them covered. Not necessarily by five different people - but covered, with clear ownership, no matter how many humans are in the room.
The Clarity Steward owns strategic coherence - the “why” behind everything the Cell does. They ensure everyone understands the vision, can connect their work to it, and that the strategy doesn’t drift or dilute as it meets reality.
The Execution Orchestrator turns vision into reality by orchestrating AI agents, automated systems, and selective human support. They own delivery - timelines, quality, tooling, and the judgment calls about what gets automated versus what needs a human touch.
The Narrative & Distribution Steward shapes how the world perceives what the Cell builds. They own the story, the brand, the channels, and the emotional connection between product and market. They don’t do “marketing things.” They make people feel something real.
The Power & Access Steward opens doors that would otherwise stay closed. They handle the highest-stakes external relationships - top-tier customers, strategic partners, capital sources. The conversations that require a human with real authority sitting across the table.
The Integrity Steward is the immune system of the organism. They own governance, incentive design, conflict resolution, and accountability enforcement. They keep the Cell honest, fairly compensated, legally sound, and culturally healthy. Without this function, every other function eventually breaks down.
If you’ve read the Jobs of the Future series, these are familiar. If you haven’t, each one has a dedicated deep-dive article you can find on my Substack.
The ideal Cell has one person per function. Five dedicated owners, each with undivided focus on their domain. But in practice, a Cell can operate with fewer people holding multiple functions. A solo founder holds all five. A triad of three covers them between its members. The functions don’t change. The distribution of who carries them does.
What matters is understanding why these five functions, and not four, or six, or ten.
Why these five
Every durable organization needs to answer five questions. Who decides what we’re doing and why? Who actually builds it? Who tells the world about it? Who brings in the relationships and resources we can’t reach on our own? And who keeps the whole thing from falling apart?
Strategy. Execution. Narrative. Access. Integrity.
Remove any one and the organism fails. Merge two into one seat and that person becomes a bottleneck with divided attention on functions that deserve undivided focus. Add more for functions that AI can handle and you’ve added humans where you don’t need them.
Five is the minimum for full coverage. Five is the maximum before you start duplicating or diluting authority.
I didn’t arrive at this number by theory. I arrived at it by studying what actually breaks. When strategy and execution live in the same person, strategy always loses because execution feels more urgent. When integrity doesn’t have a dedicated owner, small problems fester until they become existential. When narrative is an afterthought, the market never understands what you’ve built.
The flexible seats
Beyond the five core functions, a full-size Cell has room for one or two flexible positions. These are permanent members - not contractors, not part-timers - but their anchor domain depends on the mission. A deep technical architect for a technology-heavy Cell. A financial engineer for a crypto project. A regulatory expert for a heavily regulated industry.
Strict rules apply. Flexible seat holders must still meet the generalist bar. They participate fully in strategy, narrative, distribution, and governance discussions. Their specialization is an edge, not their identity. If someone can only contribute within their narrow domain, they belong on the contractor bench, not in the Cell.
This preserves cognitive diversity without silos. Everyone sees the whole picture. Everyone can challenge everyone else. No one hides behind “that’s not my department.”
The generalist requirement
This is the part that makes people uncomfortable. Every person in the Cell must be a high-level generalist.
Not a dabbler. Not someone who “knows a little about everything” in a superficial way. A generalist in the Cell means someone who has enough depth across multiple domains to direct AI systems intelligently, evaluate outputs critically, and engage meaningfully in every strategic conversation at the table.
In practice, this means every member has a primary domain they own, but they can hold their own in any other domain’s discussions. The Execution Orchestrator can challenge the Narrative Steward’s positioning. The Integrity Steward can question strategic direction. The Clarity Steward can push back on execution timelines.
When the whole Cell can pressure-test a decision from five different angles simultaneously, the decision quality is dramatically higher than anything a siloed organization can produce.
Here’s the honest truth about how this works in practice: the higher the quality of the generalists, and the higher the standard they hold each other to, the better the Cell performs. A Cell of five exceptional generalists who challenge each other relentlessly will outperform any traditional organization many times its size. A Cell of five average performers who avoid hard conversations will produce average results with a fancy structure on top.
The Cell is a multiplier. Put in great people who push each other, and it multiplies excellence. Put in mediocre people who play nice, and it multiplies mediocrity. The framework doesn’t do the work. The people do.
How accountability works
The Cell has a multi-layer accountability system. I’m going to outline the principles here and go deeper on the mechanics in a later article when I describe the full five-person Cell in operation.
Layer one: functional accountability. Each steward has final authority in their domain and full accountability for outcomes. Narrative fails? The Narrative Steward owns that. Execution slips? The Execution Orchestrator owns that. No diffusion of responsibility. No “the team decided.”
Layer two: peer accountability. Every member has both the right and the obligation to challenge any other member. This is institutionalized, not informal. Any member can trigger a full Cell review - a “red flag” pause - if they see integrity risk, strategy drift, or incentive corruption. No hierarchy can block this.
Layer three: integrity enforcement. The Integrity Steward has special powers - incentive arbitration, behavioral enforcement, and the ability to initiate removal proceedings. Their decisions are binding unless overruled by supermajority. This prevents the slow cultural decay that kills organizations over time.
And there’s a final layer most organizations are too afraid to define: removal. Grounds exist. The process is clear. Unanimous consent for admission, supermajority for removal. No ambiguity. No hostage situations.
This system works because it assumes something uncomfortable: that good people can still act badly when incentives push them wrong, when accountability is unclear, or when nobody is watching. The Cell doesn’t rely on everyone being virtuous. It creates conditions where virtue is the easiest path.
What makes this different
Most organizational frameworks in the AI discourse focus on technology integration. Microsoft’s “Frontier Firm.” McKinsey’s “agentic organization.” They all start with the question: how do we plug AI into the existing model?
The Cell starts somewhere else: what is the minimum viable human structure for judgment, accountability, and trust - and how do we build AI leverage underneath it?
The technology is a given. AI handles execution. The unsolved part is the human structure. Who decides. Who owns. Who is accountable when things go wrong. Who keeps the system honest over time.
The spectrum in practice
Most people reading this are not going to assemble a full five-person Cell tomorrow. Some of you are solo operators right now, holding all five functions yourself. Some are in partnerships or small teams of two or three. Some run larger organizations wondering how any of this applies to them.
All of those situations are valid starting points. The Cell framework scales across configurations, with each one having different strengths, limits, and failure modes.
The single player has maximum speed but no one to challenge their judgment. The triad has the first real checks and balances but members carry stretched attention across multiple functions. The full Cell of five has complete coverage with one person per function - the ideal, but also the hardest to assemble.
Over the next few articles, I’ll walk through each configuration. How it works. Where it breaks. Who it’s for.
Starting with the loneliest and most common one: the single player.
This is Part 2 of the Company of the Future series. Next: the single player - what happens when one person tries to hold all five functions, and why AI doesn’t solve the problem you think it does.