Core Concepts

Assembly is built around a simple idea: autonomy fails when coordination is implicit.

Most autonomous systems are designed as pipelines. Decisions are made upstream, execution happens downstream, and coordination logic is often hard-coded or assumed to be stable. Assembly takes a different approach by making coordination explicit.

Autonomy vs coordination

Autonomy describes the ability of a system to decide and act independently. Coordination describes how multiple decisions and actions remain coherent over time.

A system can be autonomous and still fail if its actions are poorly coordinated. As systems scale beyond a single agent or robot, coordination becomes the dominant constraint.

Intent vs instruction

An instruction specifies exactly what to do. An intent specifies what should happen.

Assembly operates on intent rather than instruction. Intent captures objectives, constraints, and priorities without prescribing a fixed execution path. This allows the system to adapt when plans become invalid.

Planning under uncertainty

Plans are not commitments. They are temporary hypotheses about how intent might be fulfilled.

Assembly assumes that plans will fail partially or completely. Replanning is treated as a normal operation rather than an exception.

Execution in imperfect environments

Execution is inherently unreliable. Actions may succeed, fail, or produce incomplete results. Assembly treats execution as a probabilistic process that must be observed and evaluated continuously.

Failure as a first-class concept

Failure is not an error state to avoid. It is a source of information.

Assembly is designed to isolate failures, prevent cascading effects, and adapt system behavior based on execution feedback rather than assumptions.