PRD outline — skeleton

Download the MarkdownSkeleton_PRD_Outline.md · 2.4 KB · from The artefact set §4

Headings and field lists only. Re-typed from the structure used on a sovereign OSINT platform built for a defence innovation challenge; no content from that or any other project.

Header block: product · MVP number and name · version · date · owner (by role) · status (draft / reviewed / approved) · links to the master spec, the eval framework and the success metrics

1 · Problem statement

• Who has the problem, in one sentence per actor
• The number that proves it, with its source, or [TBD]
• What happens today without the product

2 · Scope and non-goals

In scope (this MVP)

• One line per capability; each maps to at least one user story below

Non-goals (binding)

• What this MVP will not do, and why; a non-goal is changed only by a new PRD version
• Declined risks from the risk register land here with their reason

3 · Actors

4 · User stories with acceptance criteria

Grouped by capability. Each story carries: ID · as-a / I-want / so-that · acceptance criteria as observable checks · the NFR it depends on · the eval case it creates, if any.

Capability group A

Capability group B

Candidate stories — surfaced, not committed

• Stories heard and recorded, explicitly out of this MVP, so they are neither lost nor implied

5 · Functional requirements

6 · Non-functional requirements

Dimensions for an AI product include: latency (P50, P95) · abstention rate on unanswerable inputs · citation validity · availability in the target environment (e.g. air-gapped) · data residency · explanation coverage · resource ceiling (memory, CPU) · auditability

7 · Metrics hook

• Which KPI in the success-metrics document this MVP moves
• Which guardrail metric must not degrade
• Which eval floor this MVP must clear, by reference to the eval framework

8 · Open questions and risks

9 · Traceability appendix