Problem
Intent is repeated
Domain meaning is copied into several layers. Each copy can drift.
Cohesive Systems
Cohesive is a language and compiler family for authoring durable system meaning as an executable semantic system graph.
Building Blocks
Compose your system graph
Each Cohesive block contributes capabilities to the system graph (state transitions, relations, workflows, APIs, UIs, infrastructure, agent context, etc).
Cohesive.Core
public sealed class ShipmentShape : Shape
{
public ShipmentShape()
{
Identity(nameof(Shipment.Id));
Field<TenantId>(nameof(Shipment.TenantId));
Field<ShipmentStatus>(nameof(Shipment.Status));
}
}The problem
Software meaning is scattered
Modern systems repeat the same intent across schemas, services, APIs, interfaces, jobs, and documentation.
Over time, meaning gets scattered across files, frameworks, and conventions. AI sharpens the pressure: code is easier to generate than systems are to understand.
Problem
Structure is implicit
Behavior hides in framework conventions and local assumptions. The system's semantics are hard to inspect.
Problem
Agents infer too much
Agents reconstruct intent from implementation details instead of working from an explicit model.
The Cohesive approach
Cohesive puts the system model above code. The graph captures structure, behavior, constraints, interfaces, runtime needs, and agent context.
It is not documentation beside the system. It is compiler input.
Step 1
Define the system
Capture structure, behavior, interfaces, workflows, and runtime requirements in the graph.
Step 2
Generate the code
Compile the graph into codebases and runtime artifacts for the technologies teams already use.
Step 3
Give agents the model
Expose the graph as structured context so agents work from explicit constraints.
Graph-first
Graph-first, not code-derived
Code-derived graphs help navigation, but they are still reverse engineering. They explain code after it already exists.
Cohesive's core artifact is authored first. It carries formal semantics and generates the code and runtime artifacts.
Code-derived graph
Descriptive
Inferred graphs help humans navigate code, but they remain downstream and inherit its gaps.
Cohesive graph
Executable source of truth
An authored Cohesive graph is formal enough to compile, check constraints, and guide change.
Migration can start from existing code, but the destination is not a permanent mirror. The destination is an explicit graph that governs implementation.
A graph agents can read
Agents use the same graph at build time and runtime.
Build time
Development agents
Use the graph as structured context for generating, refactoring, and testing code from declared semantics.
Runtime
Operational agents
Use the graph to understand allowed actions, effects, permissions, evidence, approval boundaries, and recovery paths.
Semantic layer
A semantic layer above the stack
Cohesive does not replace your database, cloud, frontend, workflow engine, IDE, or agent tool. It defines the semantic layer they can read from and write back to.
The graph describes what the system is; the compiler projects it into the technologies teams already use.
Explore
Try the code
Start with the building blocks and see graph-authored semantics become code and runtime artifacts.
Direction
See the vision
Read how semantic system graphs change authoring, operations, and AI context.
Contact
Get in touch
Share what you are building or ask about applying Cohesive to an existing system.