Digital portfolio management and EU AI Act governance in one console.

Open the Inventory and add every active or candidate initiative — own work, vendor pilots, shadow tools. Record total budget and, if work is already underway, the forward-looking (remaining) budget too. Bulk-edit via TSV if you already have a list — the round-trip includes the forward-looking column.

For each initiative open Evaluate, set the four 0–100 dimension scores with confidence, add the AI Act block, dependencies and Stage-gate evidence. The Soft-Sync hint catches contradictions; the decision log keeps the audit trail.

Check the Portfolio Map (Scale / Pilot / Assist / Stop), the Value-at-Stake view and the Governance bands. Concentration and Operating Model surface where the portfolio is over-exposed; the EU AI Act register tracks per-initiative readiness.

Set the annual budget cap and Skill Catalogue (Settings → Capacity). The Prioritisation page computes a cut-line: funded vs. parked, with budget and per-skill utilisation bars. Toggle between total and forward-looking budget basis, use „Make it fit“ to promote upstream dependencies, exclude individual initiatives, or hit „Full reset“ to clear every what-if at once.

Run Scenarios for −10 % / −20 % budget, freeze a Snapshot, then export the Executive Memo and Board Pack (PDF/PPTX) — bilingual, with the full Stage-gate and AI Act readiness sections. The Board Memo and Steering KPIs surface the forward-looking budget alongside Net 3-year value so the board sees what is still allocatable.

Settings › Taxonomy gains a Stage & gate names card: rename Idea, Pilot, Scaling, In Production, On Hold, Retired and the three gates to your organisation's vocabulary. The names apply everywhere — tables, funnel, reports, exports and the methodology text — in both languages. Data values never change, so imports, backups and the API stay stable. Leave a field empty to return to the built-in name.

Settings › Organization gains an admin activity log: who granted whom access, role and permission changes, settings, profile and initiative changes — newest first, filterable by action group and free text. Deleting a user account now erases that person's data end-to-end, keeps the organisation's business records intact and puts the erasure itself on record. Under the hood this release hardens the platform with security headers, webhook rate limiting and scrubbed error reporting. Database indexes on all tenant keys keep list views fast as portfolios grow.

Gate risk now says what it actually means. A bypass is only recorded when an initiative is advanced up the ladder without completing that gate — parking it On Hold, retiring it or moving it back a stage are logged in the decision log but no longer stamped as governance failures. Evidence for a gate that has already been passed can be updated in place, so a production system's sign-off can be refreshed without walking the status backwards; completing the gate behind the current status also clears a recorded bypass. The review interval after which evidence counts as stale is now an org setting (Settings › Stage-gate evidence, default 180 days) instead of a fixed six months, and it is carried in the JSON backup. In the funnel report, the three risks finally look different: red for a bypassed gate, amber for stale evidence, blue for evidence never captured.

Everything you set in Strategic Prioritisation — budget basis, budget trade-off, per-skill FTE deltas, Make-it-fit promotions, manual excludes and queue filters — is now saved per user and survives navigation and reload, until you press Reset to baseline. Members need the new “Can edit Prioritisation” right, granted in Settings (admins always have it). The Board Memo and board-pack exports reflect your saved scenario (with a clear what-if note), and the Capacity-vs-budget KPI and Board Actions budget checks use the budget you set there.

A new Reports page builds ad-hoc tabular reports over the whole data model: pick any of ~45 fields — master data, scores, economics, governance, status-report lights — and order them freely. Group by status, business unit, priority bucket and more with automatic counts and money totals per group, on top of the shared filters and free sorting. Named reports can be saved per portfolio mode (including filters, if you choose) and the current view exports to Excel with collapsible groups.

Each initiative now carries an overall status: the worst of its latest report's four traffic lights. The Inventory gains a sortable Report status column plus a filter facet (green, yellow, red, no report) that works on every filterable page and share link. The Status Report page shows the overall light as its own column ahead of the four dimensions.

The Roadmap page now shows where your skills go: an expandable per-initiative overview with status, budget, usage and total FTE, and a stacked chart summing FTE demand per skill and month across every initiative's timeline. A dashed line marks the total capacity of the skill pool and a marker highlights the current month; both views follow the page's filters. Prioritisation also moved up in the menu, right below the Roadmap.

Initiative managers can now file dated status reports rating scope, timeline, budget and skills as green, yellow or red, each with a short comment plus an overall note. A new Status Report page shows the whole portfolio as a traffic-light grid — grouped by lifecycle status with the worst lights first, a status filter, trend arrows against the previous report and drill-down into commentary and history. The board memo, PPTX and PDF pack gained a matching Initiative Status chapter, and a new granular permission controls who may edit reports.

When a teammate saves an initiative you have open, their changes now merge into your form field by field: sections they changed glow briefly with a who-did-it note, your unsaved edits are never overwritten, and a genuine collision (same field, both changed) asks you per field — take theirs or keep yours. Read-only viewers get fully read-only screens. Background tabs pause polling and catch up on return.

Admins can now set board-level red lines under Settings → Financials & rules: a budget-concentration cap per business unit, a value floor for funded initiatives, a cap on the share of high-governance initiatives, and an EU AI Act readiness floor for production systems. The Settings card shows a live pass/fail per guardrail. Every breach also lands on the Board Actions digest as a 'decide now' item. Guardrails round-trip through full backups.

New Roadmap page: every initiative as a bar on the time axis, grouped by business unit, status or category, with a today line, dependency tooltips and overdue markers. The Board Actions digest also gained EU AI Act enforcement countdowns — incomplete high-risk documentation escalates to Decide-now as the legal date approaches.

The digest now closes the value loop for systems in production: initiatives realising well below their promised benefit become a decision (below 50% achievement is critical), and systems running with a value promise but no recorded actuals are flagged as unverifiable. New Benefits filter category.

New Board Actions page: a ranked list of the decisions the portfolio's own data calls for. Covers stop/defer candidates, stage-gate discipline, capacity and budget shortfalls, business-unit dominance, and outdated or incomplete AI Act documentation. Filter by category; every item links straight to the place to act. Rules-based and documented — no black box.

The console now runs as a multi-tenant SaaS. Each organisation is an isolated tenant; members sign in and get role-based access (admin / member / viewer) plus granular per-section permissions set in Settings. A per-field audit trail records who changed governance-critical, Core Metadata and EU AI Act fields, with created-by and last-edited-by shown on every initiative.

Producing EU AI Act documentation is now a one-click step. Every AI Act artifact, operational obligation and supporting document ships with a bundled, editable .docx template, and a new Document Templates section in Settings lets you keep the built-in template or point any row at your own external URL — complete with preview and a placeholder reference, and these choices survive backup and restore. From the Evaluate form, Create from Template downloads a ready-to-edit document with the organisation, AI system name, owner, status and last review date already filled in from the initiative, leaving every other field as guided placeholders.

Initiatives that reach In Production now steer on realised figures instead of business-case estimates. Once an in-production initiative has actuals data, its costs and benefits switch from the planned business case to the recorded actuals, while everything still in earlier stages keeps using its business-case reference — so there is zero drift between expected and realised until real numbers exist. A Soll/Ist comparison aligns expected against realised per production year, giving a clean view of how each live initiative is actually performing against plan.

The EU AI Act surface gets a major upgrade. The dedicated AI Act page is rebuilt into clear tabs with a slide-over detail panel and at-a-glance risk badges, and the Evaluate form now consolidates all AI Act inputs — role, Annex category, GPAI and transparency flags — into one grouped block. Each initiative gains an Operational Obligations checklist that surfaces only the duties its role and risk class trigger — deployer (Art. 26), high-risk provider, GPAI and systemic-risk (Art. 53/55), Art. 50 transparency and importer/distributor — each tracked with status and due date and feeding the Art. 26 deployer readiness signals. Portfolio Readiness now splits into two views: a Team view tracking implementation progress as a 0–100 steering score for internal prioritisation, and an Audit view giving each requirement a binary conformant / open / not-applicable status — an initiative is audit-ready only when every applicable requirement is conformant. Classification is sharpened with explicit Annex I and Annex III high-risk paths plus the Art. 25 GPAI value-chain.

Six capabilities land in V4.2. Named portfolio filters (#168) let users save, load and share named subsets of the portfolio with full Prioritisation funding capture so every view is immediately actionable. Cut-line exclusions (#172) let individual initiatives be flagged as excluded from the capacity cut-line — cascade warnings and dependant tooltips surface the knock-on impact. Make-it-fit (#155) adds a Zap button on every parked row that promotes the row and its transitive upstream chain above the cut-line, then iteratively raises the budget trade-off and per-skill FTE deltas by the exact gap needed (min 0.5 FTE / +5 pp, up to 40 passes) until the target is funded; a new Full reset button clears every promotion together with trade-off and FTE deltas in one click. The Board Pack PPTX (#149) is fully redesigned with a restructured content flow and refreshed slide layout. The screens and menu information architecture (#150) have been consolidated into grouped navigation, merged pages and shared headers across the product. Single-initiative export and import (#151) adds a JSON + Excel round-trip so individual initiatives can be handed off, edited offline and re-imported without a full workspace export.

The AI Act cockpit becomes operational across the product. Inventory adds Act Role + Annex III, Risk Tier + HITL and an Act-Readiness traffic light as dedicated columns, optional toggles for DPIA, Model Card, AI Literacy and EU-DB Registration, an amber stale-document indicator and saved column presets. The /ai-act page gains an AI Literacy tab (Art. 4) and a Serious Incidents tab (Art. 73) with amber/red deadline banner. Three new Steering KPI tiles cover literacy, incidents and stale docs. Settings expose per-artefact review thresholds (Art. 9/10/11/13/27/47–48/53/72) plus reminder windows. The Board Pack ships a toggleable EU AI Act Readiness section in PDF and PPTX. Import/Export adds CSV import of the incident register; the JSON backup round-trips every Act field. Demo data fills literacy, incidents and artefact maps across the AI initiatives.

A new /ai-act page lands with four tabs: Explanation (plain-language primer on scope, the four risk classes, roles, Annex III, GPAI, the seven core obligations under Art. 9–15/17, transparency under Art. 50, penalties under Art. 99 and the phased application milestones from Feb 2025 through Aug 2027), Portfolio Readiness (per-initiative scores across classification, documentation, oversight, logging and conformity), Detail (applicable-articles list per initiative with Art. 5/50/Annex-III reasoning) and Timeline. Evaluate gains a collapsible EU AI Act block — Role (Provider / Deployer / Importer / Distributor / Not-in-Scope), Annex III category, GPAI + systemic-risk flags, Art. 5 prohibited practices and Art. 50 transparency flags — plus an amber soft-sync hint that fires when the formal classification contradicts the legacy Risk Tier (e.g. Annex III set but Tier ≠ High-Risk). Demo initiatives are pre-populated with sensible AI Act data so the readiness, articles and timeline views are immediately meaningful.

Capacity is now multi-dimensional. A new per-mode Skill Catalogue in Settings holds skills with FTE caps; Evaluate gains a Required Skills section per initiative. The Prioritisation cut-line treats every skill cap as a hard constraint alongside the budget envelope — green/amber/red utilisation bars and a red "Skill bottleneck" badge on parked initiatives make the binding constraint visible. Demo data seeds a mode-specific catalogue and realistic demand; skills round-trip through JSON backup.

Cost & value economics now go beyond point estimates. Each initiative carries P10/P50/P90 ranges for benefit, an explicit Year-1 to Year-3 ramp profile and a risk-adjusted NPV (raNPV) using the configured discount rate. Portfolio totals add Value-at-Stake (P10–P90) and Σ raNPV — both tiles now appear on the Portfolio Map and the Steering KPIs cockpit, so the economic picture is consistent across surfaces. The Portfolio quadrant chart also adapts its axis domain to the actual score range, so bubbles use more of the chart area and size differences are easier to read. Fully backwards-compatible: existing initiatives keep their point estimates as P50 until ranges are entered.