apiVersion: jdai.upss/v1
kind: Vision
id: vision.jdai.product
version: 1
status: draft
metadata:
owners:
- JerrettDavis
reviewers:
- upss-vision-architect
lastReviewed: 2026-03-07
changeReason: Establish the first canonical UPSS artifact for repo-native product intent.
problemStatement: JD.AI lacks a single authoritative product-intent artifact that humans and agents can both consume and validate.
mission: Enable safe, agent-driven software delivery from repo-native specifications.
targetUsers:
- id: user.product-owner
name: Product owners
needs:
- Define product direction, scope, and measures of success in a durable repository artifact.
- id: user.delivery-agent
name: Delivery agents
needs:
- Read product intent directly from the repository and trace implementation work back to it.
- id: user.governance-reviewer
name: Governance reviewers
needs:
- Audit whether capabilities, tests, and operational changes remain aligned with approved product intent.
valueProposition:
summary: JD.AI turns the repository into the authoritative product blueprint so intent, implementation, and validation evolve together.
differentiators:
- Product intent lives beside code, tests, and deployment assets instead of in disconnected planning systems.
- Structured specs are machine-readable for agents while remaining legible to human reviewers.
- Traceability and validation can be enforced in CI to reduce specification drift.
successMetrics:
- id: metric.traceability-coverage
name: Traceability coverage
target: ">=95% of implementation artifacts map back to approved product intent."
measurement: Automated traceability audits across specs, code, tests, and deployment artifacts.
- id: metric.validation-adoption
name: Validation adoption
target: "All new canonical spec types ship with repo-native validation before general use."
measurement: CI enforcement coverage across UPSS spec directories.
constraints:
- Specifications must be stored in-repo and versioned with the product they describe.
- Every canonical spec type must support both human-readable guidance and machine-readable structure.
- Validation must run in CI without depending on external control planes.
nonGoals:
- Replace human approval for high-risk releases or security-sensitive changes.
- Encode sprint plans, milestones, or staffing decisions inside the vision layer.
trace:
upstream:
- docs/index.md
- docs/architecture/index.md
downstream:
capabilities: []