Principle
The autonomous company does not begin when the human disappears
English translation of the original Spanish publication. Original dates, authorship and documentary images are preserved.
An organization can introduce agents, automate processes and produce more without having resolved who decides what all that activity is for.
That is where our question about the autonomous company begins.
At D31, we work from a concrete tension: reality changes while the organization continues to operate. People need to interpret what is happening, coordinate and execute. When the distance between strategy and execution grows, more technology can add capacity without closing that distance.
Autonomy requires structure. A comprehensible purpose. Decisions that can be distributed. Sufficient context. Enforceable limits. Responsible people capable of making corrections.
We call 51/49 the principle that preserves human sovereignty over that whole. It does not allocate percentages of work. It expresses who retains the decision over purpose, limits, exceptions and consequences.
That sovereignty begins before an action is authorized: it needs people able to deliberate, disagree and leave. It continues during execution: it requires explicit mandates and the ability to interrupt. And it remains afterward: it requires evidence, learning and remedy when something fails.
The international discussion offers ideas against which to examine this formulation. Microsoft published a humanist code for consultation on September 14; it is not yet using it to train its models. The Cooperative AI Foundation proposes mechanisms for governing interactions among multiple agents. These are independent proposals that let us examine our questions at other scales. Microsoft · Cooperative AI Foundation.
Our proposed contribution is to assemble those questions within the transformation of real organizations. TechNúcleo™ connects capabilities and responsibilities; High Frequency Execution™ (HFE™) offers a cycle of decision, action, evidence, correction and learning. The outstanding test is to demonstrate that combination under observable conditions.
Working definition: an autonomous company can configure and reconfigure human and agentic capabilities while preserving human deliberation, authority over purpose, traceability of decisions and the ability to correct their consequences.
Autonomy is built when people can act with shared judgment and the system retains the ability to answer for what it does.
If your organization doubled its execution capacity tomorrow, could it also explain, stop and correct what it executes?
© 2026 D31. All rights reserved.
From Stated Principles to Observable Control
Microsoft's Humanist AI code describes intended behavior and acknowledges that the document does not guarantee its models' current performance. Source: Microsoft AI.
At D31, that distinction reinforces an organizational task: checking what happens when a person needs to correct course. Autonomy requires purpose, mandates and intervention routes to be understandable to participants. This is a proposed organizational application; Microsoft's document does not validate the D31 model.
The next test should examine a specific decision: who authorized it, what context it used and how it could be reviewed. Learning begins when the organization can explain that sequence.
Before adding another layer of technology, it helps to see how ready the organization is to decide, coordinate and execute.
Take the Organizational Self Check-Up™