Expect domain languages to explode. More open, the better (2/8)

In a world where context matters as much as code, the language of the domain must be expressed separately from the code in Java or Python.

A modern domain language must embrace following characteristics to be compatible with the design, build and deploy fast paradigm of AI-enabled software engineering.

Ontological definition as code. When most code may be generated, enterprise ontology is the true intellectual property - a first step towards capturing knowledge, an ever evolving construct, which makes data meaningful, captures relationships and past actions for better decision making.

Express intent within any work context, a.k.a. the reason behind the actions. This is important for explainability, of why an action has taken place, whether by human or an AI or deterministically by an automation.

The same data when presented to situations with different intent, would trigger a different choice of actions.

Express the boundary of work (a.k.a guardrails), with mathematical precision. Mathematical precision and not non-determinism is required to operate the governance framework that ensure alignment with regulatory, business monitoring and risk rules.

Enable mathematically precise testing outcomes. Insurance against regression and inviolability of governance boundaries must be demonstrated, when code gets cheaply generated.

Domain languages that support precision in testing can bring down the cost/effort of reliability, significantly.

Designed to enable deep co-ordination. Co-ordination gaps exist in the ecosystem and need solving structurally by facilitating exchange of ontological definitions, intent, expected actions and policies between systems.

When domain languages are open, they create opportunity for coordination across systems and beyond an organisation.