Gentrack AI Data Intelligence Platform
Revenue Products · Tool Elimination · Customer Uplift · Multi-Cloud — Click any card to explore with AI
📦 GLDM Current State — What Exists Today (Production)
📐 Data Model
50+
Entities in production: dims, facts, refs, SCDs. DWH + MDS_SRC schemas. Daily refresh 03:00 NZT.
LIVE👥 Customers
6
Mercury, Genesis, EA, Pulse, PWC, Senoko. NZ + AU + SG. All with Tableau dashboards live.
PROD⚙️ Matillion Jobs
185
Transformation + Initial + Stream jobs. Per-customer variants (NZ/AU/SG). Daily automated runs.
$600K/yr📊 Reports
40+
Tableau workbooks + datasources: Billing, Aged Debt, Customer Service, MDS, Transactions, Enrolment.
LIVE
🗺️ Product Plan — From Current GLDM to AI-Driven Platform
🚀 Phase 1: AI Uplift
Weeks 1-4 — Deploy AI agents alongside existing GLDM. First products: Anomaly Detection + AI Copilot. Zero disruption to current reports.
IMMEDIATE🔄 Phase 2: Replace Matillion
Weeks 5-12 — AI reads 185 Matillion JSONs → generates Polars/SQL. Side-by-side validation. Cut over customer by customer. $600K/yr saved.
8 WEEKS💰 Phase 3: Revenue Products
Weeks 8-20 — Ship 10 AI products: Churn, Leakage, Forecast, Network, 360, Tariff, Carbon, Payment, Copilot, Anomaly. $4-8M Year 2 revenue.
12 WEEKS☁️ Phase 4: Open Lake
Weeks 16-24 — Iceberg on S3. Snowflake reads via External Tables. Athena for AWS-native. BigQuery for GCP. No vendor lock-in.
6 MONTHS
🔮 New Version: GLDM v2 — Fully AI-Driven Data Intelligence
🧠 AI-Built Data Marts
User prompt → Agent discovers source → builds mart → HITL review → deploys. 1 hour vs 8 weeks. Semantic layer auto-generated.
🔧 Self-Healing Pipelines
Schema changes auto-detected. Agent proposes fix. HITL approves. Deploys automatically. No 3am PagerDuty. Monday review only.
⚡ Hours to Onboard
New customer signs → Agent connects → discovers → profiles → maps → transforms → deploys. Hours, not months.
💬 Ask in English
"Show me Mercury customers with 50% consumption drop" → AI queries GLDM → returns table + chart + recommended action. No SQL needed.
🌐 Cross-Customer Learning
Patterns from Genesis inform Mercury proactively. Federated intelligence across all customers. Collective IQ without sharing data.
📈 Revenue Prediction
"Q3 revenue will be $X based on consumption trends + churn risk + payment patterns." Board-level forecasting from the data platform.
$4-8M
AI Revenue (Y2)
10
AI Products
~$2.5M
Tools Replaced
16
AI Agents
<3mo
First Revenue
6
Customers Ready
💰 Revenue-Generating AI Products — Market-Based Pricing Estimates
Pricing based on comparable utility analytics products (Bidgely, Itron Analytics, Oracle Utilities Insights, C3.ai). Actual pricing TBD with product/commercial team.
🔍 Anomaly Detection
$30-80K
Real-time meter & billing anomaly flagging. Catches errors before customers complain.
4 WEEKS📉 Churn Prediction
$50-150K
Predict customer departures. Identify drivers. Recommend retention actions. 2-5% saved.
6 WEEKS💧 Leakage Detection
$100-300K
Find unbilled consumption, meter bypass, tariff errors. 2-4% revenue at risk recovered.
8 WEEKS📊 Demand Forecast
$50-200K
Load prediction at ICP/feeder/network level. Weather-adjusted. Reduces spot market exposure.
8 WEEKS⚡ Network Capacity
$80-250K
Predict overloads. EV impact. DER hosting. Defer $1-5M capex per avoided upgrade.
10 WEEKS🏠 Customer 360
$40-100K
Unified: billing + consumption + payments + interactions + vulnerability. Better CX.
4 WEEKS📈 Tariff Optimisation
$60-150K
Simulate tariff impacts across segments. Data-backed regulatory submissions. Protect margin.
8 WEEKS🔄 Payment Propensity
$40-120K
Predict payments. Segment debt. Optimise collection timing. 5-15% improved recovery.
6 WEEKS🌱 Carbon & ESG
$30-80K
Scope 1/2/3 from energy data. Customer carbon footprint. NZ ETS + AU NGER compliance.
6 WEEKS🤖 AI Copilot
$20-60K
Natural language: "Show ICPs with 50% consumption drop" → instant answers. No SQL needed.
4 WEEKS
🗑️ Tool Cost Elimination — Approximate (~ to be confirmed)
Qlik Replicate
~$1.5M/yr
9 customers. Replaced by CDC Engine (Snowflake Streams + keyset pagination). Already built.
BUILT ✅Talend Cloud
~$200K/yr
Genesis + Mercury migrations. Replaced by DMS 16 AI agents. Build #465 deployed.
BUILT ✅Matillion
~$600K/yr
6 customers × ~$50-100K. AI reads Matillion JSON → generates Polars/SQL. 8 weeks to parity.
8 WEEKSAWS Glue
Avoided
Proposed by UK. Not adopted. Fargate + Polars: faster, cheaper, multi-cloud capable.
AVOIDEDManual Dev Time
~$300K/yr
Developer time on SQL, ETL, reports. AI agents auto-generate transforms + data marts.
PROGRESSIVE
🏗️ Platform Architecture — AI-Native, Multi-Cloud
🧠 Control Plane
Gentrack-hosted. 16 agents + GAC + Bedrock. Orchestrates all intelligence. Multi-tenant.
⚙️ Data Plane
Customer cloud. Fargate/Cloud Run extractors + Polars transforms + CDC. $0 idle.
💾 Storage Layer
Iceberg (open) → Snowflake + Athena + BigQuery all read same data. No lock-in.
🔮 AI Engine
Bedrock (AWS) / Vertex (GCP) / Azure OpenAI. Brain + Nerves + Memory. Multi-step reasoning.
🤖 AI Agent Arsenal — 16 Autonomous Agents
🔍 DiscoveryAgent
Scans sources. Tables, FKs, row counts, schemas. 9 connectors.
📊 ProfilingAgent
Null%, cardinality, distributions, quality score. Works on any data.
🔀 TransformAgent
AI-generates Polars/SQL/dbt. Reads source mapping → outputs executable code.
🏗️ DataMartAgent
User prompt → AI builds mart. 5 min vs 8 weeks. HITL approval gate.
🔄 CDCAgent
Snowflake Streams + watermark. Replaces Qlik. Proven on 125M rows.
✅ ValidateAgent
Rules, schema drift, regulatory (NMI/MIRN). Self-healing pipeline.
⚖️ ReconcileAgent
Source vs target. 99.32% proven (Genesis BCT). Row + value + hash checks.
🎯 MasterAgent
Think → Swarm → Review. Orchestrates all agents. DAG execution.
⬆️ Customer Uplift — Zero Disruption, Pure Value-Add
⚡ Mercury (NZ)
$200-400K/yr
500K customers. 14.2B readings. Churn + anomaly + demand forecast. GLDM live.
⚡ Genesis (NZ)
$150-300K/yr
BCT done (99.32%). Audit done. Add leakage + payment AI. Immediate upsell.
🦘 PWC (AU)
$150-350K/yr
Large AU network. Network capacity + leakage + ESG (NGER). High value.
⚡ EA / Pulse (NZ)
$100-250K/yr
Competitive NZ market. Customer 360 + churn + tariff optimisation.
🌏 Horizon Power (AU)
$200-400K/yr
New customer. Remote grid challenges. Full platform deployment. Anomaly + network + forecast.
🇸🇬 Senoko (SG)
$80-150K/yr
Competitive SG market. Demand forecast + churn. Spot market exposure reduction.
📊 Return on Investment — AI Products + Tool Savings (~to be confirmed)
📈 AI Product Revenue
$4-8M (Y2)
10 AI products × 12 customers. Based on market pricing for comparable utility analytics products.
💰 Tool Savings
~$2.5M/yr
~Qlik + ~Talend + ~Matillion + dev time. Approximate — needs confirmation with procurement.
🎯 Investment
$350-550K
2-3 engineers × 6 months to build AI product suite. Platform already exists (sunk cost). Running ~$50K/yr.