Skip to main content
The contextual layer is where Spine Fabric transforms its reservoirs into actionable intelligence. It is the mechanism that answers: for this specific focus, at this moment, what is the complete working picture that humans, Twin, and agents need to reason and act? This page covers what context assembly does, what goes into an assembled context, when it runs, why on-demand construction matters, and how this layer powers both the human workbench and the Spine Agent Network.

​
What Context Assembly Does

Context assembly is the process of producing a scoped, coherent working picture given a focus and a purpose. The focus might be an account, a ticket, a contract, or any business entity. The purpose might be human review, agent invocation, workflow progression, or event interpretation. The contextual layer gathers everything relevant, trims what is not, and delivers a unified view. The result is not a raw data dump. It is a purpose-built picture where entities, timelines, evidence, and memory are arranged so that the consumer can orient immediately and act with confidence.

​
What Goes Into an Assembled Context

An assembled context draws from multiple reservoirs and layers them into a single deliverable. The composition varies by scope and trigger, but typically includes the following elements.

Focus Entity and Neighbors

The primary entity being examined, plus its directly related entities. Relationships are resolved through identity linking so that the picture is not fragmented across sources.

Recent Events

The event timeline on the focus and its neighbors. Includes status changes, communications, decisions, and system updates, ordered so the consumer sees what happened and in what sequence.

Evidence

Communications, documents, call transcripts, and artifacts attached to the focus or its recent events. Evidence is surfaced, not buried, so agents and humans can inspect source material directly.

Prior Decisions and Receipts

Outcomes from previous agent actions, workflow steps, or human decisions that bear on the current focus. Memory prevents redundant reasoning and supports continuity.
In addition to these core elements, domain overlays may be applied depending on the consumer. A support view might layer health scores and SLA status. A finance view might layer contract value and renewal risk. The fabric provides the raw picture; projections tailor it.

​
When Assembly Runs

The contextual layer is not a background batch process. It is triggered on demand at specific operational moments.
  • Workbench focus. When a user selects an entity in the Human Workbench or AI Workbench, the fabric assembles the working picture for that surface.
  • Agent invocation. When the Spine Agent Network routes a task to an agent, the network requests assembled context scoped to the agent’s capability and the task at hand.
  • Workflow step. When a workflow advances, the next step receives a fresh context capturing the state produced by prior steps.
  • Event intake. When an external event enters through a connector and triggers a rule or automation, the fabric assembles context around the event’s focus so that routing and response occur from full knowledge.

​
Why On-Demand Assembly Matters

Constructing context at read time rather than maintaining pre-materialized snapshots gives Spine Fabric three essential properties.
  1. Freshness. The assembled picture always reflects the latest intake, resolution, and state. No stale projection lingers because a scheduled job failed to run.
  2. Scope control. The same fabric can produce a narrow view for a quick agent check or a broad view for strategic review, without maintaining multiple copies.
  3. Latency discipline. Because assembly is scoped and purposeful, the fabric only fetches what the consumer needs. The cost of building a picture is proportional to its depth.
On-demand assembly means the fabric never lies about state. The picture is built from what is true right now.

​
Powering Human Workbench and Cognitive ESB

The contextual layer is the bridge between Spine Fabric and every surface that consumes it. For the Human Workbench and AI Workbench, assembly produces the entity pages, timelines, and insight panels that users interact with. For the Spine Agent Network, assembly produces the scoped context that agents receive at invocation so they do not act blindly. Without the contextual layer, the Spine Agent Network would have no shared picture to route against. Agents would lack orientation. Handoffs would break. The contextual layer is what makes the network cognitive rather than merely mechanical.