FEATURE SPEC

Questions or comments?

Post them as a comment on the tracking issue -- requires a GitHub account with access to this repo.

Natural Language Query Interface

Parent Feature Set: ../concept.md.

Note

Following information is reverse engineered for [AI-DLC]. Review and remove this note before proceeding.

Important

To follow the AI-DLC process, fill this spec with PM assistance first.

GitHub Issue Description (verbatim)

Issue #614, titled "[Reporting] Natural Language Query Interface":

Summary

Feature set

No Acceptance Criteria, Expected Business Value, or Additional Context sections were present on this issue — stating that fact rather than inventing content.

Matching PRD Content

Reproduced verbatim from ../reference/Reporting_PRD_v3_2026-07-17.docx, §3 and §5.

Nova Natural Language Reporting Interface — Reporting is the on-demand, conversational data capability of the Novaworks HR workspace... Any permissioned user types a workforce question in plain English. Nova interprets the intent, identifies the relevant tables in the Novaworks data model (PERSON, WORKER, JOB_ASSIGNMENT, LEGAL_ENTITY, COREHR_AUDIT, and related), enforces the user's permission scope, and returns results in the correct delivery modality. No query language, no filter UI, no waiting for a data team. The model does not generate or hallucinate data values — it constructs queries; the data layer executes them against live HR records.

§5 ServiceNow AI Data Explorer Integration — Because Novaworks is built natively on ServiceNow, all HR workforce data lives in ServiceNow tables (x_nova_corehcm_* and related). Rather than building a standalone natural language query engine inside Novaworks, Nova Reporting is architected as a wrapper on top of ServiceNow's AI Data Explorer — the native NLP-to-data capability within the ServiceNow platform. ServiceNow AI Data Explorer handles natural language interpretation and data retrieval against the underlying ServiceNow data layer. Novaworks contributes the HR-specific context, permission scoping, Nova conversational surface, and workforce-specific delivery logic on top.

Flow 1 — Ask: The user types a plain-language request into the Ask Nova input. If ambiguity is detected, Nova surfaces a one-sentence interpretation statement with an inline option to adjust before rendering. For unambiguous requests, Nova renders directly.

Ask Nova (navigation) — Not a separate tab — a persistent, always-visible input surface present on every page of the Reporting workspace, and available platform-wide via a sparkle button (✨) entry point. Users ask questions from anywhere, at any time.

Sources

• GitHub issue #614
• ../reference/Reporting_PRD_v3_2026-07-17.docx, §3 and §5