Zava Innovation Hub

SPFx 1.24.0 GA upgrade (October 28, 2026): Build dependencies target GA. React/ReactDOM use 18.3.1. Earlier readiness reports, screenshots, test counts, and tenant evidence remain historical, not fresh GA host certification. Restore was validated using the configured Microsoft npm proxy; npmjs.org did not yet expose all required GA packages. Use an approved registry with 1.24.0 available; do not commit credentials.

An offline innovation-management showcase with 17 registered tools, one shared SPFx bundle, five role workspaces, and intent-specific inline experiences. Copilot selects the business job; controlled React UX owns evidence, calculations, validation, review, and explicit human confirmation.

Runtime: SPFx 1.24.0, React/ReactDOM 18.3.1, React types 18.3.31/18.3.7, Heft 1.3.2. Node.js 22.14+ within 22.x, 24.x, or 26.x is supported by the manifest; this release was validated with Node 22.23.2. Preserve the pinned versions and fast-uri 3.1.8 override.

No live tenant data, automated approvals, messaging, capital transfers, or persistent workflow. Receipts are in memory until page reload or Reset demo data. They do not rewrite the immutable source catalog or update other dashboards. A pilot Go is learning, never scale authorization. Earlier August design/rehearsal artifacts describe a previous implementation and are historical only.

Screenshots

Enterprise Insights	Global participation

Idea submission	Inspection-only review queue

The 11 curated publication filenames and thumbnail metadata are retained. Supplementary captures in assets/intent-evidence.json cover every tool at standard, narrow, and dark settings; all seven submission flows have review/receipt evidence in each variant. assets/gallery-evidence.json and the supplementary report include current source hashes, image hashes, dimensions, runtime errors, and overflow checks.

The seven complete business operations

Every flow follows Draft → Review → Confirm → semantic session receipt. Review is a snapshot of the actual controlled inputs, not an invented readiness score. Edit draft returns to the same inputs; reset restores the invocation defaults. No receipt exists before the final confirmation.

Tool	Distinct work, evidence, and safeguards
SubmitInnovationIdea	Problem, affected audience, strategic theme, owner, evidence source, current/target process days. Target must improve baseline. Review calculates reduction and checks exact title overlap against the 120-idea catalog. This is not a semantic/live duplicate check.
BuildIdeaBusinessCase	Selected idea, investment, annual gross benefit, annual cost, benefit confidence, owner, and sourced assumption. Review calculates net benefit, confidence-adjusted net, first-year ROI, and adjusted payback. Non-positive adjusted benefit explicitly has no payback. No NPV or multi-year cash flow is claimed.
CelebrateInnovationImpact	Selected measured outcome, named contributor, specific contribution, measurement source, recognition channel, and editable message. Review composes the actual praise with the selected idea's measured amount. Confirmation prepares a story; it does not post it.
ReviewInnovationFunding	Gate eligibility, Approve/Adjust/Defer/Decline choice, requested USD amount, accountable owner, and conditions. Review shows available-before, new commitment, and remaining envelope. Allocation cannot exceed $800,000 or bypass the selected idea's gate; defer/decline commit $0.
GenerateInnovationBrief	Selected idea or complete catalog source ledger, audience, owner, headline, editable evidence narrative, and leadership decision ask. Source totals remain independent of narrative edits. Confirmation prepares a brief; there is no fake Send/Export action.
LaunchInnovationChallenge	Recruitment theme/region/audience, opening/closing dates, qualified submission target, owner, measurable success criteria, and impact/feasibility/inclusion weights. Requires a 7–90-day window, whole target 1–1,000, and a 100% rubric. Confirmation stages a mock launch.
ManageInnovationExperiment	Funded pilot, hypothesis, predeclared sample and reduction thresholds, baseline/observed process days, sample size, incident count, observation source, owner, and separate human Go/Pivot/Stop choice. Rules explain their recommendation. Insufficient sample or missed reduction blocks Go; any safety incident requires Stop. Confirmation records learning only.

Business-case formulas:

• Annual net = annual benefit − annual operating cost.
• Adjusted net = benefit × confidence / 100 − annual operating cost.
• First-year ROI = (annual net − one-time investment) / investment × 100.
• Adjusted payback = investment / adjusted annual net × 12, only if adjusted net is positive.

All money is USD. Mandatory long text is at least 10 characters; financial values are finite and bounded at runtime. Exported schemas deliberately avoid unsupported JSON Schema length/range keywords. User-entered evidence is not independently verified.

All 17 intent contracts

Only implemented parameters are advertised. Fixed analytical snapshots intentionally accept no filters. Unknown explicit IDs, unsupported regions, or empty personal/value filter results remain visibly no-match, publish no-match model context, and never fall back to another record or a healthy score.

Tool	Optional properties	Actual behavior / scope
SubmitInnovationIdea	message, theme	Initializes title and theme; complete opportunity intake as above.
GetMyInnovation	ideaId, theme	Filters ideas owned by the signed-in display name. Shows stage, impact, effort, and stage-specific next action; first 12 matches, with count.
BuildIdeaBusinessCase	ideaId, amount	Initializes the selected record and one-time investment.
CelebrateInnovationImpact	ideaId, message	Initializes selected outcome and editable recognition text.
GetInnovationReviewQueue	None	Screening/Business case/Pilot buckets and status filters change the actual visible records/counts. Detail inspection cannot commit decisions.
ReviewIdeaGate	ideaId	Exact selected gate record; inspect strengths/gaps, choose approve/send back/decline, enter rationale, review consequences, explicitly confirm. Only this tool owns gate decisions.
ReviewInnovationFunding	ideaId, amount, message	Initializes selected request, proposed capital, and editable rationale; capital constraints apply.
ExploreInnovationPortfolio	None	Curated illustrative horizontal funnel, impact/effort bubbles, investment mix, momentum, themes, and impact ranking. Read-only, not a generic filter workspace.
TrackInnovationValue	ideaId, theme	Selected-scope projected/measured totals and variance, with named owner ledger (first 8 matches). The separate trend is explicitly the whole program series.
GenerateInnovationBrief	ideaId	One idea or all 120 catalog records; sums the actual selected data.
GetInnovationGrowth	None	Fixed Jan–Dec 2026 series; submissions/conversion/value buttons change the chart and computed first-to-last change.
ExploreGlobalInnovation	region	Actual selected AMER/EMEA/APAC/LATAM marker and regional evidence; comparison gap is calculated from selection.
TrackInnovationBudget	None	Fixed mock bridge: $4.8m allocated − $2.6m committed − $900k spent − $500k forecast = $800k available.
GetInnovationPortfolioHealth	None	Illustrative leadership scorecard and exceptions, separate from exploratory portfolio charts. Not a live computed assessment.
LaunchInnovationChallenge	message, theme, region	Initializes title/theme/recruitment region.
ManageInnovationExperiment	ideaId, message	Initializes funded pilot and editable hypothesis; observations must be supplied by a person.
ExploreAgentCapabilities	query	Search all 16 operational capabilities and select a safe scenario/prompt preview. It never invokes or confirms an operation.

Default record is IDEA-001 where a single idea is needed, except funding (IDEA-044, gate-approved) and experiment (IDEA-057). Those defaults apply only when no ID was supplied.

Exact demonstration records

ID	Name	Useful route
IDEA-001	Smart Onboarding Journey	Screening gate, business case, recognition
IDEA-014	Customer Signal Studio	Screening review with data-owner/privacy gaps
IDEA-022	Supplier Carbon Twin	Inspect sent-back review
IDEA-026	Meeting Mood Scanner	Inspect declined review
IDEA-031	Knowledge Discovery Platform	Pending business-case gate; capital allocation is blocked
IDEA-044	Demand Forecasting Modernization	Approved review history and value variance
IDEA-057	Frontline Copilot Coach	Pilot learning and pilot gate
IDEA-063	Invoice Exception Automation	Measured $1.7m outcome and recognition

domain.ts contains 120 deterministic synthetic idea records, with these named overlays, 11 fictional demo personas, four regional aggregates, and 12 months of illustrative program history. REVIEW_SUBMISSIONS is a separate curated queue with phase-specific evidence and historical outcomes. The queue and aggregate funnel are demonstration snapshots, not inferred from the synthetic stage distribution. There is no implied durable progression between tools.

Full screen and lifecycle

The same mounted primary body continues inline and full screen: fields, selected record, step, reviewed snapshot, and receipt survive passive width/theme/mode changes. A new component invocation or changed intent/properties starts fresh. The retained React root is unmounted before awaited base teardown.

Role workspace defaults:

Demo profile	Work area	Default tool
Megan Bowen	My Innovation	Personal ideas
Johanna Lorenz	Programs & Pilots	Growth series
Diego Siciliani	Reviews & Gates	Inspection queue
Miriam Graham	Investment	Budget stewardship
Joni Sherman	Enterprise Insights	Portfolio command center

Expand keeps the invoked tool, including a submission or selected-value view, rather than replacing it with the area's default dashboard. Work-area navigation shows the other area's default. Returning to the original area restores the retained primary body and its visible model context. Compact changes spacing; Reset demo data clears receipts and resets mounted drafts. Navigation is a responsive icon rail with arrow-key support. Role names are demo context, not authorization.

The supported copilotBridge.updateModelContextAsync publishes committed visible information, no-match, review/confirmation snapshots, or receipt details. Uncommitted field edits are not published. Hidden primary drafts do not replace context for a visible secondary workspace. Host expansion/context errors are visible instead of being treated as success.

Conversation starters

config/conversation-starters.json is canonical. Each starter owns one registered primary tool; capability exploration remains exactly last.

#	Title	Starter	Expected inline component
1	Submit an idea	Submit an idea to reduce new-hire onboarding time by half.	SubmitInnovationIdea
2	Global innovation	Compare innovation participation and conversion across regions.	ExploreGlobalInnovation
3	Portfolio funnel	Show stage conversion in our innovation portfolio funnel.	ExploreInnovationPortfolio
4	Portfolio health	Show which innovation portfolio health measures are outside target.	GetInnovationPortfolioHealth
5	Approve an idea	Review Smart Onboarding Journey for gate approval.	ReviewIdeaGate
6	Explore capabilities	Explore what this agent can do.	ExploreAgentCapabilities

The existing M365 declarative-agent JSON and copilot/instruction.txt retain one-tool routing, neighbor exclusions, exact record extraction, and human confirmation boundaries. All existing 17 GUIDs and tool names are preserved.

Build, test, and evidence

npm ci
npx playwright install chromium
npm run capture:visual
npx tsc --project scripts\visual-harness\tsconfig.json
npm run build

npm run build validates identities/starters, agent icons, media provenance, routing documentation, current-source gallery evidence, production audit, and diff whitespace; runs production compilation, lint, and all discovered Jest tests; packages the solution and agent; audits generated plugin schemas and package bytes. The production audit must pass, not be bypassed.

The GA upgrade passed 64 Jest tests, compilation, lint, packaging, generated-plugin validation, and package-output validation. The full npm run build command currently fails at the production dependency audit on existing fast-uri@3.1.8, zod@3.24.1, and zod-to-json-schema@3.24.1 advisories. These findings still require a separately scoped dependency/schema migration; the audit gate remains enabled.

DOM tests use act from React and one retained root per test host. Coverage includes each distinct operation, measured thresholds, actual input-to-receipt values, no-match, information property effects, human gate choices, fresh invocations, teardown, and theme/mode retention.

• Curated publication: assets/gallery-evidence.json (11 thumbnails).
• Supplementary full-intent coverage: assets/intent-evidence.json and assets/intent-visuals/.
• Package contents/hashes: assets/release-evidence.json.
• Exact local validation commands/results: assets/local-validation.json.
• Deployable solution: sharepoint/solution/zava-innovation-portfolio.sppkg.
• M365 package: teams/zava-innovation-hub.zip.

The harness supports intent, mode=fullscreen, theme=dark, ideaId, message, amount, region, strategicTheme, and query. test=1 exposes passive theme/mode controls and hidden context evidence. It is a local development harness, never a production route.

Demo guides and external gates

• 10-minute business demo
• 5-minute technical demo
• 4-minute keynote
• Designer review
• Routing matrix
• Current completion checklist and historical plan

Portraits are packaged byte-identically from the established demo reference set; names and case-sensitive paths are in src/shared/media.ts and assets/asset-provenance.json. Unknown names use initials. No profile-photo request is made. The four original design-reference PNGs are not implementation evidence. Public portrait redistribution rights still require approval.

Local deterministic/browser tests do not certify a Microsoft 365 tenant. GA tenant installation, actual natural-language routing and extraction, host CSP/bridge support, iframe focus, screen readers, and full accessibility rehearsal remain external release gates. Previous Workbench evidence must not be presented as fresh GA proof. Production identity/permissions, live data, finance, notifications, durable workflow, independent evidence verification, and media rights are intentionally outside this offline sample.

Version history

Version	Date	Comments
Unreleased	October 28, 2026	Upgrade to SPFx 1.24.0 GA; React/ReactDOM 18.3.1 and Heft 1.3.2.
1.0.1	October 22, 2026	Restore the sample-specific README view tracker required by gallery validation.
1.0	October 20, 2026	SPFx 1.24 RC0 showcase and validation update.