Aptiv Types & Composition
An Aptiv is the atomic unit of Wantware — not a function, not a container, not a service. It is a modular structure of executable meaning, defined by Meaning Coordinates, that encodes what a behavior is for as well as what it does. Every [.wv] stream is composed entirely of Aptivs.
The eight Aptiv types
Aptivs are organized into eight families, each handling a distinct domain of behavior. The same Meaning Coordinate framework governs all eight — they differ in what they represent, not in how they are structured or validated.
| Type | Role |
|---|---|
| PowerAptiv | Executes model logic, control flows, and signal transformations across platforms. The primary execution layer — handles algorithms, services, engines, and hardware-level instructions. |
| MindAptiv | Represents goals, planning strategies, user context, and reflective reasoning. Used to model intent, decision criteria, and AI-like inference within a governed structure. |
| RecordAptiv | Manages structured data, datasets, telemetry, and evolving logs tied to model inputs or outputs. Replaces traditional databases and file-based data stores. |
| SignalAptiv | Ingests and processes real-time sensor inputs — video, audio, LiDAR, X-ray, and other continuous data streams. Signal data becomes semantically structured rather than raw bytes. |
| StoryAptiv | Sequences behavior over time with causality, chronology, and outcome tracking. Used for workflows, event modeling, simulation sequences, and adaptive learning paths. |
| ExperienceAptiv | Adapts interfaces, interactions, and feedback based on user role, device context, or AI-generated predictions. Platform-agnostic — a single ExperienceAptiv renders correctly across mobile, desktop, and XR. |
| BeliefAptiv | Models probabilistic logic, thresholds, uncertainty, and conditional trust parameters. Used wherever behavior needs to be governed by confidence levels or policy-bounded inference. |
| IdeaAptiv | Defines and evolves conceptual frameworks that guide permissible behavior and policy adaptation. The governance layer for what a system is allowed to know, do, or become over time. |
All Aptivs are organized into 64 categories (8 families × 8 subsystems), creating a shared grammar for combining and scaling behaviors across the platform. This structure is what allows Aptivs from different sources to compose without middleware.
PowerAptivs — encapsulating existing code
PowerAptivs are the execution core of the Essence stack. They are also the primary mechanism for incorporating legacy code, third-party libraries, and AI models into a [.wv] stream without rewriting them.
Rather than replacing existing logic, PowerAptivs wrap it — adding Meaning Coordinates, trust enforcement, and semantic traceability around code that continues to run as-is internally. From the outside, a PowerAptiv is indistinguishable from a codeless Aptiv: it exposes the same interface, participates in the same composition model, and is subject to the same policy validation.
What PowerAptivs encapsulate
- Platform-specific binaries — DLL, dylib, .so — wrapped with declared intent and trust constraints
- Cross-platform codebases — C/C++, Python, Lua, Rust, Prolog — ingested via Elevate
- Compilers and runtimes — interpreted and compiled language runtimes, JIT environments
- AI and ML models — wrapped with policy and output constraints, adding traceability and override controls
- Security controls — encryption libraries, licensing, hardening routines, unit test harnesses
- Structural metadata — changelogs, version lineage, proofs of correctness
Protected Aptivs — trust without exposure
Not all Aptivs need to expose their internals to participate in the trust model. Protected Aptivs allow proprietary logic or sensitive data to remain concealed while still being subject to Essence's validation framework.
The mechanism works through declared intent: even a protected Aptiv must express what it is for using Meaning Coordinates. Essence evaluates behavioral outcomes against that declared intent without accessing the protected internals. If behavior diverges from declaration, SecuriSync flags or blocks it — the same as any other Aptiv.
Protected Aptiv properties
- Internal logic — concealed; not inspectable by other Aptivs or system components
- Declared intent — required; must be expressed via Meaning Coordinates
- Trust enforcement — applied; behavior is validated against declared intent at runtime
- Composition — full; protected Aptivs compose with other Aptivs without special handling
- Regulatory compliance — supported; protected Aptivs can satisfy audit requirements without source disclosure
How Aptivs compose
Aptivs combine into Wantverses — structured collections that coordinate multiple Aptiv types to serve a unified purpose. A Wantverse is not a bundle or a package in the conventional sense. It is a live, policy-governed system whose behavior emerges from the interaction of its Aptivs rather than from a fixed execution path.
Because all Aptivs share the same Meaning Coordinate foundation, composition requires no middleware, no glue code, and no API contracts. An ExperienceAptiv and a SignalAptiv from entirely different sources can be combined in a [.wv] stream and will interoperate correctly — their shared semantic structure is the interface.
Aptivs in AI governance
When AI models are introduced into Wantware they enter as PowerAptivs, not as opaque black boxes. Each of the eight Aptiv types plays a distinct role in governing how those models behave at runtime.
| Aptiv type | Role when governing AI behavior |
|---|---|
| PowerAptiv | Executes model logic and controls inference flows across platforms |
| MindAptiv | Represents goals, planning strategies, and reflective reasoning about model outputs |
| RecordAptiv | Manages structured data, datasets, and evolving logs tied to model inputs or outputs |
| SignalAptiv | Ingests and structures real-time sensor inputs before they reach the model |
| StoryAptiv | Sequences AI behavior over time with causality and outcome tracking |
| ExperienceAptiv | Adapts interfaces and feedback based on AI-generated context or predictions |
| BeliefAptiv | Models probabilistic logic, uncertainty thresholds, and conditional trust parameters |
| IdeaAptiv | Defines and evolves the conceptual frameworks that bound permissible AI use |
Aptivs replace functions, services, scripts, and containers as the unit of composition in Wantware. All eight types share the same structural foundation, which means they compose without middleware and are subject to the same trust enforcement regardless of whether they contain legacy code, AI models, or codeless meaning. PowerAptivs are the path for incorporating existing systems — wrapping rather than replacing them.