Nishi vs the Field -- Office Suite

How this is scored. Every Nishi cell is measured against the real organ source on disk (present = the emit symbol exists; absent = it does not exist anywhere). Competitor cells = documented 2025/2026 capability from the cited office-sota bar. Unlike Computational Math, this domain scores honestly LOW: the Nishi office suite is a sovereign foundation, not a SOTA product. The liar-kill enforces coverage under 600 and that the marquee gaps are genuinely absent.

Capability	Nishi	M365 Copilot	Workspace Gemini	LibreOffice	Notes
Word processor
.docx (OOXML) read/write	Yes	Yes	Yes	Yes	nx_docx -- Word-openable OPC package
.odt (ODF) read/write	Yes	Part	Part	Yes	nx_odt; cloud leaders import/convert
Rich formatting (styles/bold/fonts)	Yes	Yes	Yes	Yes	nx_docx run-properties (w:rPr)
Tables	Yes	Yes	Yes	Yes	nx_docx w:tblPr
Images / media	No	Yes	Yes	Yes	no w:drawing yet
Page layout (headers/sections)	No	Yes	Yes	Yes	no sectPr
Track changes / revisions	No	Yes	Yes	Yes	no w:ins (doc-level versioning differs, see below)
Comments / annotations	No	Yes	Yes	Yes	no commentReference
TOC / footnotes	No	Yes	Yes	Yes	no footnoteReference
Spell / grammar	No	Yes	Yes	Yes	no organ
Mail merge	Yes	Yes	Part	Yes	nx_mailmerge {{field}} -> per-record .docx
Spreadsheet
.xlsx read/write	Yes	Yes	Yes	Yes	nx_xlsx SpreadsheetML
Cell typing	Yes	Yes	Yes	Yes	numeric vs inlineStr
Formulas / functions	No	Yes	Yes	Yes	no formula cells -- values only (the core gap)
Charts	No	Yes	Yes	Yes	none
Pivot tables	No	Yes	Yes	Yes	none
Conditional formatting	No	Yes	Yes	Yes	none
Macros / scripting	No	Yes	Part	Yes	none
Presentation
Slides author (.pptx/.odp)	No	Yes	Yes	Yes	NO presentation organ
Collaboration
Real-time co-editing	No	Yes	Yes	Part	no OT/CRDT -- the biggest cloud gap
Sharing / comments / permissions	No	Yes	Yes	Part	nx_vault ACL exists but unwired
Version history	Best	Part	Part	Part	EXCEEDS: tamper-evident, content-addressed, never-lose -- cloud history is mutable/deletable
In-app AI (the 2025/2026 headline)
In-doc AI draft/rewrite/summarize	No	Yes	Yes	No	Copilot in Word / Gemini in Docs
AI data analysis in sheets	No	Yes	Yes	No	Copilot in Excel / Gemini fill
AI presentation generation	No	Yes	Yes	No	Copilot in PowerPoint / Gemini in Slides
Agentic multi-step	No	Yes	Part	No	Copilot Agent Mode GA 2026-04; gated on our LLM arc
Storage / documents / sovereignty
Document management (versioned store)	Yes	Yes	Yes	Part	nx_vizsla_doc store/get/history
Doc generation from structured data	Yes	Yes	Yes	Part	nx_vizsla_docgen fields -> .docx
Records / estate workflow	Yes	Part	Part	No	nx_estate_inventory never-lose
Cloud / offsite sync	No	Yes	Yes	Part	local seg_store only; oauth2 roadmap
Sovereign: on-device, no telemetry	Yes	No	No	Yes	PARITY with LibreOffice; cloud leaders send data up
Open format / no lock-in	Yes	Part	Part	Yes	both OOXML and ODF

Honest verdict. NOT a state-of-the-art office suite — a sovereign foundation with correct .docx/.xlsx/.odt interop, mail-merge, and versioned document management, plus one genuine exceeds (tamper-evident content-addressed never-lose version history, which even LibreOffice lacks) and privacy/open-format parity with the open-source SOTA. The 2025/2026 leaders' differentiators are all absent: agentic in-app AI (Copilot Agent Mode / Gemini), presentations entirely, deep spreadsheet (formulas/charts/pivots), and real-time collaboration. Coverage is a capability fraction, not a rank.

Generated by nx_swcompare_office (sovereign NishiLang organ) — every Nishi cell verified against real organ source on disk; competitor cells from office-sota-bar-2026-07-08.txt. Zero JS, zero trackers.