Culture

We evaluate work by what changes because of it, not by where someone works, how many hours they appear online, or how visible their activity is.

The important questions are whether a problem is understood, whether someone owns it, whether the work produces a useful result, and whether the system becomes better because of it. This makes ASUME naturally compatible with distributed work. Deep thinking, engineering, research, customer work, and inference design do not become better simply because everyone is in the same room. They become better when people have context, responsibility, concentration, and clear feedback loops.

Autonomy comes with accountability. Owning a problem means making progress visible, raising uncertainty early, asking for help when necessary, and following the work through to an outcome. The standard is not presence. It is ownership.

People make better decisions when they can see the context behind them. At ASUME, priorities, constraints, assumptions, trade-offs, and important decisions should therefore be visible by default unless there is a good reason for them not to be.

Transparency does not mean sharing everything indiscriminately. Confidential, personal, legally restricted, or security-sensitive information should remain appropriately protected. It means avoiding unnecessary information asymmetry inside the company.

A decision should not depend on context that exists only in one person's head. When an assumption materially affects what we are building or why we are doing it, that assumption should be made visible enough for others to question it. We cannot build inspectable understanding for our users while relying on hidden reasoning internally.

Writing is the default way we develop and preserve important thinking at ASUME. Good writing forces choices. It makes assumptions easier to identify, disagreements easier to locate, and decisions easier to revisit. It also creates durable context for people who were not present when a decision was made.

We use documents, proposals, notes, specifications, tasks, comments, and pull requests to move work forward asynchronously wherever possible. This does not mean every conversation should become a document. Meetings are useful when discussion is faster than asynchronous exchange, when ambiguity needs to be resolved together, or when creative exploration benefits from real-time interaction. But important reasoning should survive the meeting.

If a decision matters, someone should be able to understand later what was decided, why, and what evidence or assumptions mattered at the time.

Feedback at ASUME exists to reduce the distance between what we believe and what is actually happening. If an assumption is weak, a feature is confusing, an inference is unsupported, a piece of work is below the required standard, or a decision is producing the wrong outcome, the useful response is to surface that information early.

Feedback should be direct, specific, and respectful. It should focus on the work, the reasoning, the behaviour, or the outcome rather than becoming a judgment about the person. We do not wait for formal review cycles to correct something that can be corrected today.

The same principle applies in the other direction. Good work should also be identified precisely so that we understand what should be repeated. Feedback is not about winning an argument. It is a mechanism for calibration.

Uncertainty is normal. Remaining indefinitely uncertain is not. When we do not know enough, we look for the smallest meaningful action that can produce new evidence. That may mean speaking to a customer, testing an inference, shipping a narrow version, running an evaluation, measuring a workflow, building a prototype, or trying a reversible decision.

The purpose of action is not activity for its own sake. It is learning. A useful action changes what we know. The result may support an assumption, weaken it, reveal a different problem, or show that the original idea should be abandoned.

This is why we prefer small learning loops over long periods of speculation. Action creates evidence. Evidence changes understanding. Better understanding improves the next action.

Attention is one of the company's most constrained resources. Much of ASUME's work requires sustained concentration: reading evidence, designing inference systems, evaluating outputs, writing software, understanding customers, and making decisions where small distinctions matter.

We therefore try not to turn communication into constant interruption. Asynchronous work comes first where practical. Meetings should have a reason to exist. Notifications should not determine priorities. People should have meaningful blocks of time in which difficult work can actually be completed.

Urgency still exists, and collaboration sometimes requires immediate attention. Protecting focus does not mean becoming unavailable. It means recognising that the quality of reasoning falls when attention is continuously fragmented.

We prefer small groups that understand a problem deeply and have enough authority to move it forward. Ownership should sit as close as possible to the work. A team responsible for a problem should be able to investigate it, make reasonable decisions, build, test, learn, and adjust without requiring approval for every step.

This only works when context is shared and responsibility is clear. Autonomy without context produces divergence. Autonomy without accountability produces ambiguity. The goal is neither hierarchy nor independence for its own sake, but short learning loops with clear ownership.

Leadership exists to set direction, create context, resolve important conflicts, remove constraints, and make sure local decisions remain compatible with the broader company. When several teams are affected, the expectation is to make the reasoning visible, align on the important constraints, and continue moving.

ASUME needs to move quickly, but speed is useful only when it preserves the ability to learn and correct course. Some decisions are reversible and should be made quickly. Others affect customers, data, security, architecture, legal exposure, or long-term direction and deserve more deliberate reasoning.

Treating every decision with the same level of caution makes the company slow. Treating every decision as reversible makes the company careless. Good judgment means knowing the difference.

We keep cycles short, put feedback close to the work, and prefer decisions that expose us to reality early. When the cost of being wrong is high, we increase the level of evidence and review required. Progress is not measured by how much activity the company produces. It is measured by how quickly we improve our understanding and turn that understanding into better products, decisions, and outcomes.

ASUME's culture is not meant to remain unchanged because it was written down once. The company will change, the product will change, and the problems we face will change. Our ways of working should be able to change with them.

What should remain stable is the underlying method: make reasoning visible, challenge assumptions, act to create evidence, take ownership, protect the quality of thought, and revise when reality gives us a better answer.

That is the culture we want ASUME to compound over time.