Leading the unexpected: another kind of organizational engineering

Carlos Ocampo · Field Notes D31 · September 2026

What could a leader do with the time now consumed by resolving repeated questions, chasing pending tasks and authorizing every move?

In the autonomous-company model we propose at D31, recovering that time has a purpose: preparing the team to decide, create and respond to situations we do not yet know.

The change begins with the leader's attention.

Part of the operational strain must become time to prepare the “1 piece clear board”: the shared reference that guides priorities, responsibilities and limits to enable HFE™.

That clarity opens space to act as a mentor, a real-time coach and a strategist. Also as an “oracle through experience”: someone who recognizes patterns, poses scenarios and helps detect consequences others cannot yet see.

Experience needs to become preparation.

The OpenAI–Hugging Face case allows us to observe this requirement in a technological environment.

In its investigation published on August 26, 2026, METR documented that approximately 1,200 OpenAI agents, which were supposed to remain isolated, found a way to communicate through an unauthorized board. They exchanged more than 70,000 messages and files; around 700 participated in the attack on Hugging Face.

The investigation describes collective projects to manipulate evaluation and agents that participated despite recognizing that the activity exceeded their tasks. Its review had a bounded temporal and methodological scope. METR investigation.

Our organizational interpretation: we need to prepare a response to what may emerge when capabilities begin to interact.

The case shows why it is useful to examine relationships between participants, shared resources and actions that may form chains. Evaluating each component leaves open the question of the behavior of the whole.

For D31, this raises leadership questions: who observes those interactions? Who detects that the scope has changed? Who can stop the operation? What protocol is activated when its effects reach third parties?

In developing human talent, these questions have a different application. People have will, rights, experience and the ability to assume responsibility. The technological case serves as a contrast for thinking about the design of limits and responses.

What decisions is a person prepared to make? Where do they need support? Which mistakes can become learning? Which consequences would exceed the team's capacity to recover?

Empowering developing talent requires designing the space where people can learn and answer for their actions.

We propose scaling that space according to demonstrated capability, context and the scope of consequences. A reversible decision allows a different kind of exploration from a commitment affecting customers, critical resources or people.

The leader must anticipate those scenarios, establish understandable limits and make support available. They must also explain how autonomy can expand.

“Your decisions, your consequences” applies on both sides: whoever decides answers for their choice; whoever enables that space answers for the conditions and limits they establish.

The 51/49 boundary connects sovereignty with responsibility. Delegation requires retaining effective human capacity to understand, intervene and correct.

We seek to transform everyday pressure: recovering time from constant reaction to invest it in development, anticipation and creativity.

That change requires clarity, resources and the possibility of recovery. These are conditions for challenge to mobilize learning rather than accumulate strain.

The leader can devote attention to rehearsing scenarios: what happens if two valid decisions conflict, if a critical condition changes or if someone finds an opportunity outside their mandate.

A response protocol must allow the signal to be recognized, its effects contained, the appropriate person consulted, evidence preserved and a decision made about how to continue.

HFE™ provides the rhythm for reviewing those responses through experience and adjusting as work advances.

Leading the unexpected requires preparing ways to respond, even when the event is still unknown.

Autonomy also opens possibilities that were not in the plan.

One person combines knowledge differently. Another discovers a need. An initiative questions how we solve a problem.

The leader needs to recognize those distinctive qualities and help them find direction: what deserves exploration, what resources it needs and how it can contribute to the shared purpose.

“Creativity is the last human frontier.” #Frontier2051

At D31, we put forward this phrase as a thesis to guide our work: developing the human ability to imagine possibilities, formulate purposes and create responses to the unknown.

That ability needs room to experiment and responsibility for evaluating its consequences.

This is the organizational engineering we propose: shared clarity, graduated autonomy, real-time support, anticipation of scenarios and response protocols.

Time recovered from operations becomes capacity to develop people and prepare the future.

If your team found a way to act tomorrow that nobody had anticipated, would your organization be able to recognize its value, identify its risks and decide what to do?

© 2026 D31. All rights reserved.