The idea
What gives a domain concept its meaning?
An account is more than a name and a balance field. Its meaning also depends on what may happen, under which conditions, and which facts change its state. “Co-arising” names this attention to concepts and their relationships together.
Start with DDD · Try the relationship exercise · Keep the decision human
Keep the rule close to its consequences
A useful model helps people agree on behavior and helps engineers find where that behavior is implemented.
What have we actually agreed?
A diagram can name an account without settling whether a repeated initialization is accepted, rejected or ignored. Conditions and outcomes need their own explicit declarations.
What else must we revisit?
A rule change may affect a command, event, state fold or caller. Recorded relationships give reviewers a starting point for impact analysis, alongside code inspection and tests.
The aim is not to replace discussion with a compiler. It is to give decisions a reviewable home, derive supported implementations from them, and record where those implementations came from.
Build on DDD’s consistency boundaries
This is a way to put DDD relationships to work, not a claim to replace DDD or to define its only correct interpretation.
Tactical: connect the command, the rule and the state
An aggregate is a consistency boundary. In the supported event-sourced model, commands are checked by declared guards, accepted commands produce events through bindings, and events evolve state through folds. A slice brings that behavior together for review.
- A guard gives a declared rule an enforcement point; an unreferenced rule deserves investigation.
- A binding explains where an event payload comes from.
- A state fold explains how that event changes state. Some events may intentionally leave state unchanged.
In the bank source, CreditBalance owns account behavior such as InitializeAccount and ReserveFunds. Order and installment behavior belong to FundOrder and SipPlan, not to CreditBalance. The recorded 149 rules are across the bank aggregates, not one account aggregate F-09 · 2026-11-11.
Strategic: make context and ownership explicit
Bounded contexts, subdomains and context maps help express where a meaning applies and how teams or systems depend on one another. The language records context, subdomain and relation notes, including context dependencies and aggregate-to-subdomain relationships.
Names and types identify concepts. Behavior, conditions and consistency relationships explain how those concepts work together. The compiler checks the supported declarations; people still decide whether the model fits the business.
Feel it: the aggregate dissolve / arise demo
A relationship exercise over the exported CreditBalance subgraph. Remove or restore edges using the diagram or the keyboard-accessible checklist. It measures declared guard coverage, not execution of the bank rules. All edits stay in this exercise; they do not change the source tree or the Explorer database.
Guard edges here are expanded from slice declarations. Losing one means incomplete coverage in this exercise. The separate raw event-producer query checks for missing produces edges, not missing guards; it does not receive these local edits.
Relations are not exclusive to this approach
Ontologies and semantic platforms can model relationships, constraints and actions too. The distinction here is the engineering workflow, not “types versus meaning.”
Connect enterprise data and decisions
Semantic platforms, including Palantir Ontology, bring concepts, links and actions together around enterprise data. Their suitability depends on the platform, integration and governance needs of a team—not on a claim that ontologies lack relations.
Review files, derive code, trace changes
The truth-source workflow starts from versioned author files. It compiles declared relations and supported behavior, produces CQRS/ES code, and connects generated locations back to knowledge through source anchors.
A file-first source, explicit generation boundaries and inspectable provenance may fit your engineering workflow. They do not establish feature parity or compatibility with another platform. See the current interfaces and proposed projections.
Checks inform a decision; they do not approve it
Agents draft, humans decide
People and agents can investigate relationships and propose changes. A human confirms the business expectation and reviews the evidence. Record that decision through the project’s actual review and provenance process; a green gate is not approval.
Change the right source
The truth tree is the knowledge-editing surface. Explicit technical bindings and the handwritten zone are separate inputs with separate responsibilities. Validate the changed layer, then test the behavior against independently agreed expectations.
A name for the way we work
“Co-arising” is an invitation to consider a concept together with its conditions and relationships. In engineering terms: keep the business statement, its declared behavior and its implementation connected as they evolve.
Keep discussion visible
Files can be reviewed and diffed. That does not remove the need for discovery, examples or judgment.
Know what was checked
Validation, compilation, tests and anchors check different properties. Follow their responsibilities.
Keep useful standards
Adopt standards for their jobs. A proposed export is not a shipped integration. Read the projection status.