Pipeline functionality

The AISP pipeline provides a controlled journey from functional requirement to validated, packaged EBX development output.

Pipeline at a glance

The pipeline is deliberately staged. Each stage narrows ambiguity, adds EBX-specific context, and improves output quality before the final code package is delivered.

Review the stages ↓

From prompt to packaged output

User defines the functional requirement

The user describes the required behavior in natural language, using EBX data model context and optional reusable prompt tools to add approved guidance.

Requirement analysis

AISP converts the requirement into structured signals such as domain terms, capabilities, normalized intent, and constraints.

Component resolution

The system identifies candidate EBX component types and confidence levels, including multi-component solutions when a requirement spans more than one responsibility.

Knowledge and guidance retrieval

Version-aware EBX knowledge, project standards, and relevant guidance are selected to keep generation grounded and focused.

Prompt composition

The pipeline builds a structured, constraint-driven instruction set for the AI model rather than relying on an unstructured prompt.

Code generation and validation

Code is generated, reviewed, and refined for EBX API fit, lifecycle correctness, structure, readability, and performance.

Output packaging

Validated classes are separated into files, placed into package paths, and prepared as governed development artifacts.

The value is reflected in the consistency and quality of the results. AISP controls the journey from requirement to deliverable, rather than relying on a single unconstrained AI response..