The Autonomous Enterprise · AE Suite · Module 07

Strategy Dashboard
The Unified Intelligence View
— All eight modules. One screen.

The Strategy Dashboard is the CTO and CFO's window into the Autonomous Enterprise. It writes nothing, decides nothing, and infers nothing — it reads the operational outputs of all seven preceding modules and presents them as a unified intelligence layer. The dashboard is only meaningful because the seven modules beneath it are running. It is the capstone, not the foundation.

BigQuery · Materialised Views Looker Studio · 5 panels Read-only · No HITL · No ML C-suite · CTO · CFO · Real-time Platform ART · H3 · PI-8 · Minimal Risk
System Context — C4 Level 1

Seven module datasets in. One C-suite view out.

The Strategy Dashboard's dependency on all seven operational modules is its defining architectural characteristic. It cannot be deployed before every module it reads from is live — and the value of each additional module going live is immediately visible in the dashboard. The Strategy Dashboard is the measure of how complete the AE Suite deployment is.

Five Dashboard Panels — Source Datasets and Refresh Cadence
PanelBigQuery materialised viewSource datasetsRefreshWhat the CTO sees
Pipeline Health sd_pipeline_health ae_ccai · ae_contracts · ae_financial 15 min Inbound inquiries → qualified leads → contracted → revenue recognised. Funnel with counts and conversion rates per stage.
Revenue Recognition sd_revrec_posture ae_financial · ae_audit.hitl_events 5 min ASC 606 classification breakdown (SALE / MULTI-ELEMENT / SERVICE) · HITL-04 queue depth · override rate trend · month-to-date recognised revenue.
Fleet Intelligence sd_fleet_intelligence ae_devices · ae_audit.hitl_events 1 hour 12,000-unit RUL heatmap by region and risk tier. Fleet anomaly rate. Active HITL-06 and HITL-07 counts. Units in critical RUL zone (<14 days).
Financial Risk sd_finrisk_posture ae_finrisk · ae_audit.hitl_events 5 min 30-day anomaly alert history by severity tier. Open HITL-08 count. Average resolution time. Alert volume trend vs 90-day baseline.
Compliance Posture sd_compliance_posture ae_audit.hitl_events · ae_governance · Vertex AI Model Registry 1 hour EU AI Act scorecard: all HITL checkpoints green/amber/red by SLA completion rate. Model Card currency (days since last review). Data quality scores by source.
Architecture — BigQuery Aggregation Layer

Materialised views. Not live queries. Not a data warehouse migration.

The Strategy Dashboard does not query the operational ae_ datasets directly in Looker Studio. Each panel reads from a named BigQuery materialised view that is refreshed on a schedule. This isolates dashboard query load from operational dataset write performance, enables query result caching, and gives each panel a clearly documented source of truth that can be audited independently.

Agent State Machine

Refresh. Serve. Alert if SLA breached. Nothing else.

The Strategy Dashboard has the simplest state machine in the suite. It is not an agent in the ADK sense — it is a scheduled refresh process and a Looker Studio rendering layer. The only non-trivial state transition is the compliance alert: if any HITL checkpoint's 7-day SLA completion rate drops below 90%, the compliance panel turns amber automatically.

Data Flow — Monday Morning Session

CTO opens the dashboard. Five panels load. The week begins.

The data flow for the Strategy Dashboard is unusual — it is the CTO's browser session, not a data pipeline event. The sequence shows which BigQuery materialised views are queried in which order as the five panels render, and what the CTO sees in each panel for a representative Monday morning in a full H3 deployment.

Dashboard Mockup — Five Panels

What the CTO sees. Every module visible. Every risk surfaced.

The Strategy Dashboard mockup shows all five panels in a representative Monday morning state — healthy pipeline, normal RevRec queue, one critical fleet unit (MCH-0042, already familiar from five preceding module demos), no open financial anomalies, and a green EU AI Act compliance scorecard.

Architecture Decision Records

Two Strategy Dashboard decisions. Both about what not to build.

ADR-SD01 — Strategy Dashboard specific
Looker Studio over a custom React dashboard for C-suite rendering
A custom React dashboard was evaluated — it would give complete control over panel layout, colour scheme, interactivity, and drill-through behaviour. Rejected for three reasons: (1) Looker Studio is natively integrated with BigQuery — data source connections, refresh scheduling, and IAM are managed through GCP's existing access control framework. A custom React dashboard requires a separate backend API, a separate authentication layer, and a separate deployment pipeline. For a read-only C-suite dashboard, that operational overhead is unjustified. (2) The CTO and CFO don't need a custom UI — they need accurate, up-to-date data in a familiar, navigable format. Looker Studio's BigQuery connector handles the data freshness, caching, and rendering. The portfolio's custom React and HTML work is better spent on the HITL interfaces where the UX directly affects operational decisions. (3) Looker Studio's sharing model (share a link, view in browser, no login required for internal GCP users via organisation authentication) matches exactly how a C-suite dashboard is consumed — in a meeting, on a shared screen, by people who will never open the developer console. A custom React app adds a deployment and maintenance burden for zero user experience benefit at the executive audience level.
Accepted · Phase Platform Design · Strategy Dashboard module
ADR-SD02 — Strategy Dashboard specific
Materialised BigQuery views over live queries — refresh cadence as a design constraint
Running live BigQuery queries directly from Looker Studio on every dashboard load was the initial design. Rejected because: (1) The ae_ operational datasets are write-heavy during peak processing windows — RevRec AI classifications arrive continuously, Asset IQ RUL predictions run daily at 02:00 UTC, FinRisk Sentinel streams financial events in real time. A live query from Looker Studio during a write-heavy window competes with operational writes for BigQuery slot capacity and increases query costs. Materialised views are pre-computed and served from cache — the dashboard query load is isolated from operational write load. (2) The five dashboard panels have different freshness requirements that are not well-served by a uniform "always live" approach. The compliance posture panel refreshes hourly — re-computing the HITL SLA completion rates across the full ae_audit.hitl_events table (potentially millions of rows) on every Looker Studio page load would be prohibitively expensive. A materialised view computed hourly costs one BigQuery query per hour regardless of how many times the dashboard is opened. (3) The materialised view definitions are the documented specification of what the dashboard shows — they are queryable, auditable, and testable independently of Looker Studio. If an executive questions a figure in the dashboard, the answer is a SELECT from the named materialised view, not an explanation of what query Looker Studio ran.
Accepted · Phase Platform Design · Strategy Dashboard module
Stakeholder Rebuttals

Four objections. Each with an architectural answer.

CTO · S-01
Why five separate views — why not one unified data model?
"Five separate materialised views means five separate refresh schedules, five separate query definitions, and five separate places where the dashboard can diverge from the operational data. Why not one unified ae_strategy_dashboard table that all panels read from?"
Architectural response
A unified table would require a single refresh that aggregates all seven source datasets simultaneously — at the fastest panel's refresh cadence (5 minutes for RevRec and FinRisk). That means running the fleet intelligence aggregation (a 12,000-unit scan across ae_devices) every 5 minutes instead of every hour. That's a 12× increase in BigQuery compute cost for the fleet aggregation with no value — the CTO doesn't need RUL data refreshed every 5 minutes. Per-panel materialised views with per-panel refresh cadences match the actual data freshness requirements of each panel to the cost of refreshing it. The fleet panel refreshes hourly because hourly is sufficient. The compliance panel refreshes hourly because HITL SLA rates don't change minute-to-minute. The RevRec panel refreshes every 5 minutes because the Finance team uses it to monitor the live HITL queue during month-end close. Five views is the correct granularity — one view per business domain, each with a refresh cadence that matches the domain's operational tempo.
Evidence: ADR-SD02 (materialised views with per-panel cadence) · Panel source table (refresh rates range from 5min to 1h) · BigQuery cost optimisation (fleet scan 1h not 5min)
CFO · S-03
How fresh is the data — what's the worst-case lag?
"If I open the Strategy Dashboard at 08:45 on a Monday morning and the fleet intelligence panel last refreshed at 08:00, I'm looking at 45-minute-old data for a panel that might show a critical unit needing service. What's the worst-case data lag across all five panels?"
Architectural response
The worst-case data lag is clearly documented per panel: Pipeline Health is max 15 minutes stale. Revenue Recognition and FinRisk Posture are max 5 minutes stale — these are the panels where near-real-time freshness matters. Fleet Intelligence and Compliance Posture are max 1 hour stale. For fleet intelligence, 1-hour-old data is operationally acceptable — Asset IQ's daily RUL batch job runs at 02:00 UTC, and the HITL-06 queue that the Field Service Manager is working from is in Firestore (real-time), not in the Strategy Dashboard. The CTO uses the fleet panel for situational awareness and trend monitoring, not for operational dispatch decisions. If a critical fleet event occurs during the 1-hour refresh window, the Asset IQ HITL-06 notification reaches the Field Service Manager directly via the HITL queue — the Strategy Dashboard is not in that alert path. Each panel shows its last refresh timestamp so the CTO always knows exactly how fresh the data is.
Evidence: Panel source table (refresh cadence per panel) · Asset IQ HITL-06 (direct notification to FSM, not via dashboard) · Dashboard mockup (last refresh timestamp visible per panel)
Enterprise Architect · S-08
What happens to the dashboard when a source module is down or quarantined?
"If the Data Governance module quarantines APAC-East data and Asset IQ stops predicting for those units, what does the fleet intelligence panel show? Does it show stale data, missing data, or an error? And does the compliance posture panel alert on the quarantine?"
Architectural response
The fleet intelligence materialised view includes a data_quality_flag field from ae_governance — units with stale features (due to quarantine) are shown with a grey data_unavailable indicator in the RUL heatmap rather than a green/amber/red status. The panel caption notes "N units: data pending reinstatement" when the quarantine count is non-zero. Stale data is not presented as current data — the view query joins ae_devices with ae_governance.validation_log and flags any unit whose most recent feature record has a quality_score below 0.60 or a status of QUARANTINED. The compliance posture panel does show the quarantine event — the Data Governance section of the compliance scorecard shows the current quarantine count and the data steward resolution status. This gives the CTO visibility into the quarantine event without implying that operational decisions are being made on quarantined data.
Evidence: sd_fleet_intelligence view (data_quality_flag join with ae_governance) · sd_compliance_posture view (Data Governance quarantine count) · Data Governance FSM (stale features → confidence degradation → HITL routing)
CCO · S-02
Does the compliance posture panel constitute EU AI Act documentation in itself?
"If a regulator asks for EU AI Act compliance documentation, can I show them the Strategy Dashboard compliance panel? Or is it an operational view that summarises the real documentation — the HITL records, the Model Cards, the SHAP explanations — which lives elsewhere?"
Architectural response
The compliance posture panel is a summary view — it is not the documentation itself. It answers "is the compliance posture healthy right now?" with a green/amber/red indicator and HITL completion rates. The actual EU AI Act documentation that a regulator needs is: the HITL audit records in Firestore and BigQuery (Article 14), the SHAP explanations in BigQuery (Article 13), the Model Cards in Vertex AI Model Registry (Article 11), and the Data Governance lineage records (Article 11 training data documentation). These are all queryable directly from BigQuery or the Vertex AI console — the Strategy Dashboard is the summary indicator that tells the CCO whether all of those underlying records are being generated correctly. Think of the compliance panel as a dashboard warning light: it tells you when something needs attention. The service manual (the actual documentation) is in BigQuery. Showing a regulator the dashboard compliance panel without the underlying records would be like showing a car's dashboard to a mechanic instead of the service history.
Evidence: sd_compliance_posture view (aggregates HITL rates from ae_audit — the underlying records are the documentation) · BigQuery HITL audit records · Vertex AI Model Cards · Data Governance lineage chain (§02 of M-08)
Demo Pathway

Three minutes. Five panels. The whole suite in one screen.

The Strategy Dashboard demo is the portfolio's capstone moment — it shows that all eight modules are running, producing data, and presenting a unified intelligence view. The München unit (MCH-0042) appears in the fleet panel, linking back to ContractGuard, RevRec AI, and Asset IQ. The compliance panel shows green across all eleven HITL checkpoints.

00
Setup · 30s before
Open the BigQuery ae_strategy_dashboard dataset — show the five materialised views
Open BigQuery, navigate to ae_strategy_dashboard. Show the five materialised views: sd_pipeline_health, sd_revrec_posture, sd_fleet_intelligence, sd_finrisk_posture, sd_compliance_posture. Each has a last_refresh timestamp. This establishes that the dashboard reads from pre-computed views, not live queries.
BigQuery ae_strategy_dashboard5 materialised views
01
Open dashboard · 0:00
Open the Looker Studio dashboard — all five panels render
Open the Strategy Dashboard URL in Looker Studio. Watch all five panels render in approximately 3 seconds from the cached materialised views. Point out the last refresh timestamps on each panel. The fleet panel shows the München unit MCH-0042 highlighted in the critical tier — the same unit from three previous module demos.
"This is the view the CTO opens on Monday morning. Pipeline health from CCAI and ContractGuard. Revenue recognition posture from RevRec AI. Fleet intelligence from Asset IQ. Financial risk from FinRisk Sentinel. Compliance posture from every module. All eight modules. One screen. Three seconds to load."
Looker StudioBigQuery materialised views5 panels rendered
02
Fleet drill-through · 0:45
Click MCH-0042 in the fleet panel — drill through to Asset IQ
Click the MCH-0042 unit in the fleet intelligence panel. Looker Studio runs a drill-through query against ae_devices for that unit. Show the unit detail: 8.4 days RUL, confidence 0.78, HITL-06 open, FSM SLA 5h 47m remaining. Point out the link to the Asset IQ module page — the same unit whose contract was processed by ContractGuard and whose revenue was recognised by RevRec AI.
"MCH-0042. We saw this unit in ContractGuard when its purchase contract was analysed. We saw it in RevRec AI when the revenue was classified as MULTI-ELEMENT. Now it appears in the fleet panel because Asset IQ has flagged it for preventive maintenance. The Strategy Dashboard is where the full story becomes visible — the contract, the revenue, the asset, all connected."
Looker drill-throughae_devices unit detailAsset IQ HITL-06 link
03
Simulate compliance amber · 1:30
Show what the compliance panel looks like when a HITL SLA slips
In the demo environment, temporarily update the sd_compliance_posture view to simulate HITL-04 SLA completion dropping to 85% (between 80–90% threshold). Show the compliance panel turning amber. Show the HITL-04 indicator changing from green to amber. Point out the drill-through link to the RevRec AI HITL queue.
"If HITL-04 SLA completion drops to 85%, the compliance panel turns amber automatically. The CTO doesn't need to check the RevRec AI queue — the dashboard tells them something needs attention. They click through to the queue and can see which classifications are approaching their SLA. Reset back to green — everything healthy."
sd_compliance_posture updateAmber alert triggerHITL queue drill-through
04
Compliance posture · 2:00
Show the full compliance scorecard — all eleven HITL checkpoints
Open the compliance posture panel in full. Show all eleven HITL checkpoint indicators — all green, all above 90% SLA completion. Show the model card currency section (5/5 current), the data quality average (0.93), and the carbon savings line (18.4 kgCO₂ MTD from GreenOps). This is the EU AI Act posture in one view.
"Eleven HITL checkpoints. Five model cards. Data quality above threshold. Carbon savings tracked. This is what 'EU AI Act compliant' looks like as a live operational dashboard. Not a document filed once a year — a live view of the compliance posture, updated hourly, always queryable."
sd_compliance_postureAll 11 HITL indicatorsModel card registryGreenOps ESG
05
Query the view · 2:30
Show the materialised view SQL behind the pipeline health panel
Open BigQuery. SELECT * FROM ae_strategy_dashboard.sd_pipeline_health LIMIT 5. Show the query result — same numbers as the dashboard. Then show the view definition (CREATE MATERIALIZED VIEW) to demonstrate that it is a standard BigQuery SQL query over ae_ccai, ae_contracts, and ae_financial. The dashboard is not magic — it is a named, documented, queryable view.
"The dashboard is a view. Anyone with BigQuery read access can run the same query. The compliance posture panel the CCO sees is the same data a regulator can query from BigQuery. The Strategy Dashboard is the UI — the data is in BigQuery, open, auditable, and permanent."
BigQuery SELECT from materialised viewView definition SQL
AE Suite Complete
Eight modules. One coherent system.
The Autonomous Enterprise.

The Strategy Dashboard is the final module. Every module page follows the same seven-section template, draws on the same München account narrative, and produces the same SHAP-before-HITL audit pattern. The portfolio design is complete.

AE Suite Index
All 8 Modules →
Start
Page 01 — The Autonomous Enterprise →