Reporting — Nova Natural Language Reporting¶
Questions or comments?
Post them as a comment on the tracking issue -- requires a GitHub account with access to this repo.
Novaworks AI | wraps ServiceNow AI Data Explorer; reads x_nova_corehcm_* and related tables (see Open Questions on prefix)
Note
Following information is reverse engineered for [AI-DLC]. Review and remove this note before proceeding.
GitHub Issue Description (verbatim)¶
Issue #404, titled "[Reporting]":
Summary¶
Natural Language Query Interface
Major Change History¶
| Version | Date | Author / Notes |
|---|---|---|
| 1 | 2026-08-11 | Seeded from GitHub issue #404 and Reporting_PRD_v3_2026-07-17.docx — reverse-engineered, not yet reviewed by a PM. |
Business case¶
Reproduced verbatim from the source PRD's §2 (Problem Statement).
Organizations running on modern HR platforms are data-rich but insight-poor. Workforce data exists — in worker records, job assignments, position tables, compensation structures, compliance filings, and audit logs — but getting a meaningful answer out of it requires navigating static report catalogs, submitting requests to a data or IT team, or building and maintaining manual spreadsheets. None of these approaches scale, and none of them reflect how the people who need the data actually think about their questions.
The problem spans every role that touches workforce data. Executives ask questions in plain language and wait days for an analyst to prepare a response. Managers have no real-time view of their own team without asking HR. HR Operations specialists run reports manually, export to Excel, and reformat for each audience — a cycle that repeats every quarter with no institutional carryover. Compliance teams must remember to run the right report after each triggering event rather than being notified proactively when a threshold is crossed. Finance and department leaders who need workforce data for planning get point-in-time snapshots that are stale by the time they arrive.
The traditional solution — a static report library of pre-built reports — has been explicitly removed from scope. Research and product team review confirmed that practitioners rarely use a static library beyond an initial browse; they always need something slightly different. A library is a legacy construct built for platforms that could not interpret intent. Novaworks answers questions directly.
The PRD does not separately break out "value to Novaworks" or a "Novaworks lens" framing distinct from the above — unable to find that content in the PRD.
Goal¶
Reproduced verbatim from the source PRD's §3 opening description.
Reporting is the on-demand, conversational data capability of the Novaworks HR workspace. It is distinct from Dashboards (pre-built, role-gated OOTB views), from Insights (proactive Nova-surfaced patterns), and from Compliance & Filings (a separate, dedicated workspace with its own navigation entry point). Reporting is reactive and self-directed: the user asks, Nova answers.
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.
Scope¶
Reproduced verbatim from the source PRD's §3 (Scope: MVP vs. Phase 2).
In Scope — MVP:
- Conversational data request via plain-language text input, scoped to user's permitted field-level and BU access
- AI interpretation confirmation before rendering ambiguous requests
- View rendering: bar chart, line chart, table, summary card — with persistent data source, time range, and segment labels
- Saved Queries — Novaworks stores query definition and parameters; re-executes against live data on user request
- Time lens and segment lens adjustments before re-running a saved query
- Follow-up refinement in same conversational thread without navigating away
- Event-triggered reports: threshold-based detection, proactive notification, pre-populated Canvas view
- Scheduled Reports: Nova-generated reports delivered ahead of compliance deadlines
- Explicit plain-language data gap and restricted data messaging; no partial views rendered
- Unfulfillable request logging to data gap backlog (feeds roadmap)
Phase 2 / Out of Scope — Explicitly Removed:
- Pre-built report library / OOTB report catalog — removed. Practitioners ask questions; they do not browse a library.
- Pinned workspace with auto-refresh — AI Data Explorer does not support pinning. Replaced by Saved Queries.
- Blended quantitative + qualitative views — Phase 2, pending qualitative data source confirmation (OQ-03)
- PDF / formatted export — Phase 2
- Mobile responsive view (read-only) — Phase 2
- Scheduled / automated delivery to email, Slack, or calendar — Phase 3
- External data source federation — Phase 3
- Acquisition cohort and location segment lenses — Phase 2
- Qualitative data blending — Phase 2
Sources¶
- GitHub issue #404 and its 5 linked sub-issues (#614–#618).
reference/Reporting_PRD_v3_2026-07-17.docx— PRD v2.0, dated 2026-07-17, status "Draft — Revised per July 1, 2026 PRD Review and July 2026 architectural decisions."
Related PRDs¶
The PRD explicitly distinguishes Reporting from three adjacent surfaces,
verbatim: "distinct from Dashboards (pre-built, role-gated OOTB views),
from Insights (proactive Nova-surfaced patterns), and from Compliance &
Filings (a separate, dedicated workspace with its own navigation entry
point)." Compliance & Filings is specs-corehr's compliance-filings
Feature Set (GitHub #448) — a cross-repo reference, not a dependency; the
PRD is explicit that these are separate surfaces. "Insights" likely
corresponds to GitHub #10 (Nova Insights, out of scope for this bootstrap
per featureset-list.md).
Who this is for¶
Reproduced verbatim from the source PRD's §2 persona table.
| Persona / Role | Friction today | Value gained |
|---|---|---|
| Executive / C-Suite (CHRO, CFO, CEO) | Receives manually compiled, point-in-time snapshots that depend on one person's capacity to produce. Asks questions in plain language and waits days for an analyst response. | On-demand answers to plain-language questions in seconds. Consistent, live cross-department views that do not depend on one person to produce. Transparent data lineage on every view received. |
| Manager | Has no self-serve view of team data. Must request headcount, tenure, attrition, or milestone data from HR Ops and wait for a manual pull. | Plain-language questions about their team answered and scoped to only their data. Proactive Nova nudges when team-level thresholds are approached. Mobile-accessible team summaries (Phase 2). |
| HR Operations Specialist | Runs reports manually, exports to Excel, and reformats for each stakeholder audience. Must manually initiate reports after each HR triggering event. | Event-triggered reports delivered automatically when a threshold is crossed — no manual initiation. Instant answers to ad hoc requests via natural language. Scheduled reports delivered ahead of compliance deadlines without HR Ops initiating them. |
| Finance & Department Leader | Receives periodic HR snapshots that are stale on arrival and cannot be further segmented without requesting a new extract. No self-serve access to workforce data scoped to their department. | Self-serve plain-language queries scoped to their department. Views that can be reshaped by time period or segment without involving HR Ops. |
| HR Administrator / System Admin | Report taxonomy is inconsistent across modules. Event-triggered reports must be manually initiated. No runtime enforcement of role-based data access at query time. | Configurable Reporting Taxonomy consistent and discoverable across the platform. Event trigger library configurable without engineering involvement. Role-based data access enforced at query execution time. |
What we are building¶
One Feature folder per linked GitHub sub-issue. The PRD's "Three-Mode Reporting Engine" names Routine (Scheduled) Reports as a third mode alongside On-Demand and Event-Triggered — no GitHub sub-issue maps to Scheduled Reports specifically (see Open Questions).
| # | Feature | Folder |
|---|---|---|
| 1 | Natural Language Query Interface (#614) | natural-language-query-interface/spec.md |
| 2 | View Rendering (#615) | view-rendering/spec.md |
| 3 | Saved Queries (#616) | saved-queries/spec.md |
| 4 | Time and Segment Lenses (#617) | time-segment-lenses/spec.md |
| 5 | Event-Triggered Reports (#618) | event-triggered-reports/spec.md |
Data Model (Conceptual)¶
No conceptual entity/relationship diagram is given anywhere in the source
material — unable to find that content in the PRD. The PRD does name
specific platform tables Nova Reporting reads from, verbatim: "PERSON,
WORKER, JOB_ASSIGNMENT, LEGAL_ENTITY, COREHR_AUDIT, and related" — same
table set named in specs-corehr's compliance-filings/concept.md. It also names a
persistence table for Saved Queries: "query definition stored in
x_nova_corehcm_saved_report or equivalent" — not diagrammed here since
no relationships to other entities are described.
Table-Stakes & Differentiation¶
Unable to find explicit competitor names or table-stakes framing in the
PRD. It does frame the architectural differentiation directly, verbatim:
"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... Novaworks contributes the HR-specific
context, permission scoping, Nova conversational surface, and
workforce-specific delivery logic on top." Reproduced in full in
natural-language-query-interface/spec.md.
UX & Interaction Design¶
Reproduced verbatim from the source PRD's §4.
Link to Figma: "[URL — required before PRD moves to Approved status | Owner: Claire Partlow / Aaron]"
Reporting is a desktop-primary experience within the Novaworks platform, accessible from a designated Reporting entry point in the Novaworks navigation bar. The workspace is organized into three sections: Ask Nova (persistent input), Saved Queries / My Workspace (saved queries and role-gated pre-built views), and Scheduled Reports. Compliance & Filings is a completely separate workspace with its own navigation entry point — it is not a section or tab within Reporting.
Six key user flows are documented in the PRD (Ask; See; Save; Re-run a
Saved Query; Event-Triggered Delivery; Scheduled Report Delivery) —
reproduced per-Feature in each sub-issue's spec.md where they map to a
specific Feature, rather than repeated in full here.
Open Questions¶
| # | Question | Blocks |
|---|---|---|
| 1 | The PRD's own Figma link is a placeholder: "[URL — required before PRD moves to Approved status]" — no design artifact exists yet. | UX & Interaction Design accuracy for every Feature under this set. |
| 2 | The PRD's §5 states the underlying tables are x_nova_corehcm_* — the same prefix specs-corehr/CLAUDE.md already flags as stale against worker-document-management's and critical-core-hr-transactions' confirmed x_novaw_corehcm_*. This PRD adds a third data point using the older x_nova_corehcm_* spelling. |
Scope section accuracy; whether naming-and-references.md needs correcting. |
| 3 | No GitHub sub-issue maps to "Routine (Scheduled) Reports," the third mode named in the PRD's Three-Mode Reporting Engine (alongside On-Demand and Event-Triggered) — is it folded into Event-Triggered Reports (#618), or missing a sub-issue of its own? | Whether a 6th Feature needs seeding. |
| 4 | The PRD's own Open Questions are referenced by ID (OQ-03, "pending qualitative data source confirmation") but the PRD excerpt available doesn't include a full Open Questions section listing OQ-01 through OQ-03 in full — only this one inline reference was found. | Completeness of this concept's own Open Questions relative to the PRD's. |
| 5 | This entire concept was seeded from a one-line GitHub issue summary and one PRD, with no PM review yet — every section above needs a PM pass before this is treated as ready for Construction. | Everything downstream of this file. |
For internal engineering review only. Not for external distribution.
Novaworks AI | Confidential