The ZORBA Manifesto¶

The Enterprise Has a New Workforce. It Doesn't Have a New Architecture.¶

Something fundamental has changed in how organisations operate, and the frameworks we rely on haven't caught up.

For decades, enterprise architecture has been built on a single assumption: that work is done by humans. Processes are designed for humans. Governance structures oversee humans. Org charts map human reporting lines. Capability models describe what humans can do.

That assumption is now wrong.

AI agents — autonomous, capable, increasingly sophisticated — are entering the workforce not as tools but as participants. They don't just assist; they decide, execute, escalate, and collaborate. They hold context. They take ownership of outcomes. They operate at speeds and scales that human-only architectures were never designed to accommodate.

And yet, the frameworks we use to describe and govern enterprise operations remain stubbornly, exclusively human.

ZORBA exists because this gap is dangerous.

The Bolt-On Fallacy¶

The instinct of most organisations — and most framework authors — has been to bolt AI onto existing structures. Take a process framework. Add "AI-assisted" to a few steps. Publish a whitepaper. Move on.

This is the equivalent of describing the internet as "a faster fax machine."

When you bolt AI agents onto a human-centric architecture, you get:

Invisible agents — doing critical work with no architectural representation
Ungoverned autonomy — agents making decisions outside any authority framework
Accountability gaps — outcomes that no human or agent is clearly responsible for
Collaboration chaos — no shared language for how humans and agents work together
Trust by default — agents operating at autonomy levels nobody explicitly chose

The bolt-on approach doesn't just miss an opportunity. It creates risk. Real, material, enterprise-level risk.

The Defining Question¶

Every enterprise in the next decade will confront the same architectural question:

Where does the human end and the agent begin?

This is not a technology question. It is an architectural question. It touches strategy, governance, operations, culture, risk, and value creation simultaneously.

Which decisions require human judgement, and which can be delegated to agents?
At what point does an agent escalate, and to whom?
How do you audit a decision chain that includes both human intuition and algorithmic reasoning?
What does "accountability" mean when the executor is not a person?
How do you design processes that are legible to both humans and machines?

These questions cannot be answered by technology teams alone. They cannot be answered by adding a chatbot to an existing process map. They require a new architecture — one that treats the human/agent boundary as a first-class design concern at every layer of the enterprise.

What ZORBA Is¶

ZORBA — the Zontally Reference Business Architecture — is that architecture.

The name is no accident. Zorba is a Greek name and term meaning "to live each day with passion" — representing a lively, free-spirited, and passionate approach to life. We chose it deliberately. Enterprise architecture has too long been the domain of lifeless taxonomies and static process diagrams. ZORBA is built on the conviction that how an organisation operates should be designed with energy, purpose, and ambition — not merely documented and filed away. An architecture worth having is one that people (and agents) actually want to use.

It is an open reference model for how organisations structure themselves from strategy to execution in a blended human/AI workforce. It provides:

A layered model (Strategy → Objectives → Capabilities → Processes → Activities → Work) where every layer explicitly defines what humans do, what agents do, how they collaborate, and what governance applies.
A workforce composition model that recognises different layers of the enterprise have fundamentally different human/agent ratios — and that this composition is itself an architectural decision.
A taxonomy of agent participation — not just "AI-assisted" as a vague label, but specific, reusable patterns for how agents operate within enterprise functions: as autonomous executors, as co-pilots alongside humans, as specialists called upon for bounded tasks, as orchestrators coordinating complex workflows.
A governance and trust framework that addresses the unique challenges of governing a blended workforce: agent accountability, decision authority matrices, escalation protocols, audit trails, and the principle that human override is always possible but not always necessary.
A common language that executives, operators, architects, technologists, and AI agents themselves can use to reason about enterprise operations. The same model, the same vocabulary, the same architecture — regardless of whether the participant is carbon or silicon.

What ZORBA Is Not¶

ZORBA is not a technology specification. It does not prescribe which AI platform to use, which models to deploy, or which integration patterns to follow. Technology choices change. Architecture endures.

ZORBA is not a replacement for existing frameworks. TOGAF, APQC, ITIL, SAFe — these address specific concerns and ZORBA is designed to complement them. But none of them answer the question ZORBA answers: how do humans and agents co-exist as a unified workforce within enterprise architecture?

ZORBA is not an AI strategy. It is the architectural foundation that makes any AI strategy executable.

Our Thesis¶

We believe the following to be true:

AI agents are workforce participants, not tools. The distinction between a tool and an agent is autonomy. Tools are operated. Agents operate. Enterprise architecture must reflect this.
The human/agent boundary is an architectural decision. It is not an accident, not an implementation detail, not something to figure out later. It is a design choice with strategic, operational, and governance implications. It must be made deliberately, at every layer of the enterprise.
Governance of agents is non-negotiable. Autonomy without governance is chaos. Every agent operating within an enterprise must exist within a trust framework that defines its authority, its constraints, its escalation paths, and its accountability.
Architecture must be legible to agents. If your operating model cannot be parsed, interpreted, and acted upon by an AI agent, it is incomplete. The era of architecture-as-PDF-for-humans-only is over.
Workforce composition is the new org design. The question is no longer just "who reports to whom?" but "what is the optimal blend of human and agent capability for this function, at this layer, for this organisation?"
This is not the future. This is now. Organisations are already operating with blended workforces. They're just doing it without architecture, without governance, and without a shared language. ZORBA provides all three.

The Invitation¶

ZORBA is published by Zontally as an open reference architecture. We believe that the challenge of the agentic enterprise is too important, too urgent, and too cross-cutting to be solved by any single vendor or consultancy.

We invite enterprise architects, technology leaders, operations executives, AI practitioners, and yes — AI agents — to engage with this framework, challenge it, extend it, and help it evolve.

The architecture of the enterprise is being rewritten. The question is whether it's rewritten deliberately — or by accident.

ZORBA is the deliberate version.

Next: Framework Architecture →