PLATFORM TECHNOLOGY

Architecture that keeps finance in motion.
No middleware maze. No planning lag.

Nordite is a purpose-built multidimensional planning platform for the access patterns finance and operations demand: large models, frequent scenario changes, real-time collaboration, governed agents, and connected operational data.

Why we started from zero

Financial operations demand unique access patterns: simultaneous reads across many dimensions, recalculation of dependent formula chains, and real-time multi-user editing of the same planning model. Generic analytics stacks were not designed around those workflows.

We asked: what would a planning engine look like if it were designed today around live data, governed AI, scenario branching, and finance-grade auditability?

Nordite brings deep experience in enterprise planning, data infrastructure, and financial workflows into one product foundation. The result is a platform built around business models, governance, and operational responsiveness rather than disconnected technical layers.

Architecture pillars

Purpose-built planning model

The data model understands dimensions, hierarchies, formulas, versions, and scenarios directly, so finance teams can model the business instead of translating every workflow through disconnected tools.

Responsive recalculation

Assumption changes, formulas, and rollups are designed to stay interactive as models grow, keeping planning sessions conversational instead of queue-driven.

Efficient model storage

The engine is designed for sparse planning data, time series, hierarchies, and cross-dimensional references, so large models remain practical to branch, review, and govern.

AI-native from day one

The engine was designed from the start for AI agents that build, populate, analyse, and report on financial models autonomously. A machine-readable API surface that agents navigate as naturally as a human navigates a spreadsheet. Humans remain in control - approvals, policies, and audit trails govern every action - but the architecture treats AI as a first-class citizen.

Git-like branching

Scenarios work like git branches for planning - creating one is instant and zero-cost regardless of model size because only the changes are stored, not a full copy. Merge, diff, and rollback are first-class operations.

Real-time multi-user

Cell-level locking with optimistic concurrency control. Multiple users can edit the same model simultaneously without conflicts. Changes propagate instantly through WebSocket connections - no polling, no stale data, no "save and refresh" workflows.

What this means in practice

Built to keep planning work interactive as model size, collaboration, and analytics requirements grow.

Interactive
Cell edits and dependent recalculations are designed to stay responsive during planning sessions
Fast branching
Teams can create, compare, and review scenario branches without full model copies
Large models
High-dimensional planning spaces remain manageable through sparse model design
Live updates
Operational data and user edits can flow into the model without breaking the planning rhythm
Collaborative
Multi-user planning is governed with conflict handling, permissions, and audit trails
Linked logic
Formula chains and rollups update through the model so teams can see downstream effects

How Nordite compares

General-purpose infrastructure adds constraints to planning platforms: model structure has to be optimised for the database, deployments are tied to specific cloud providers, and analytical capabilities sit in separate systems. Nordite takes a different approach.

Traditional EPM
Nordite
Foundation
Relational DB / OLAP cube
Purpose-built engine
Runtime
JVM / .NET with GC
Purpose-built runtime
Recalculation
Seconds to minutes
Interactive updates
Data freshness
Batch loads (nightly/weekly)
Live streaming into the model
Scenario creation
Copy entire model (minutes)
Fast scenario branches
Dimension limits
Typically 8-12
High-dimensional models
Concurrent users
Queue-based, lock contention
Governed collaboration
AI integration
Chatbot add-on
Governed agent actions
Cell scale
Millions (with tuning)
Sparse planning spaces
API throughput
100s req/s
Responsive live updates
Dependencies
DB + cache + queue + runtime
Self-contained and sovereign
Cloud dependencies
Cloud DB + CDN + auth services
None - ships self-contained
Model optimization
Consultants restructure for performance
Model mirrors business logic
Agent performance
Batch processing, queue delays
Real-time, millisecond response
Built-in intelligence
Third-party add-ons
Native optimization, forecasting, simulation
State-based modeling
Static assumptions
State-based forecasting

What the architecture unlocks

The point of the architecture is not benchmark theatre. It is to give finance teams a platform that stays responsive, governable, and connected as planning gets more ambitious.

Self-contained
Sovereign by design
Less infrastructure orchestration before finance can start planning.
Git-like branching
Scenario branching
Teams can spin up alternatives, compare them, and merge decisions back without duplicating the whole model.
Live recalculation
No queue culture
Assumption changes ripple through models quickly enough for planning to stay conversational and iterative.
Open data edge
Lakehouse-friendly
Parquet and Arrow previews can be explored in place, so finance sees operational detail without another ETL project.
AI-native surface
Governed execution
Agents operate through explicit skills, scoped tools, approvals, and audit trails instead of side-channel automation.
One model
Shared operating context
Forecasting, optimization, simulation, reporting, and approvals all work from the same planning backbone.
Reverse stress testing
Risk quantified
Work backwards from failure thresholds to identify exactly which assumptions would break your plan.
Digital twins
Simulate before you commit
Maintain a live virtual replica of your business. Test decisions against real data before they reach the P&L.
Audit-ready by default
Built-in governance
Every model change, agent action, and data write is logged with who, what, and when. Compliance is structural, not bolted on.
State-aware
State-based intelligence
State-based models help teams understand customer lifecycle, pipeline conversion, and changing revenue patterns.
Live data feeds
Real-time business observability
Live transactional data from CRM, ERP, POS, and market feeds flows directly into the model. No batch loads, no overnight imports. Plans reflect what is happening now.
Data connectors
Continuous data integration
Connect finance, sales, operations, market, and file-based sources. Agents can monitor approved signals and surface model updates as conditions change.

Data sovereignty for regulated industries

Financial services, healthcare, defense, and government organisations often cannot use cloud-only planning tools. Nordite deploys on your infrastructure, keeping every byte of financial data under your control.

Self-hosted deployment

Run Nordite on your own infrastructure with Docker or Kubernetes. Your data stays in your environment, your jurisdiction, your network. No external dependencies required.

Air-gapped operation

The engine runs completely self-contained. Self-hosted deployments support offline mode with automatic sync when connectivity returns. No internet connection required for core planning operations.

Full data residency

Choose where your data lives. Your cloud, your data centre, your jurisdiction. Meet regulatory requirements for financial services, defense, healthcare, and government without compromise.

Responsive planning at scale

Large planning models should not force finance into slow refresh cycles. Nordite is designed to keep recalculation, simulation, forecasting, and collaboration responsive on standard deployment environments.

Efficient computation

Planning calculations are optimized for dimensional models, formulas, and rollups so teams can work interactively even as complexity grows.

Elastic analytics capacity

Deployments can be sized for heavier simulation, optimization, and forecasting workloads when teams need more analytical throughput.

Built-in analytics

Simulation, optimization, statistical forecasting, and state-based analysis work inside the planning platform. Users and governed agents can analyze the same model rather than exporting data to separate tools.

Why independence from external systems matters

When the engine is self-contained, model structure doesn't have to be optimised for a database, deployments aren't tied to specific cloud providers, and AI capabilities work natively within the platform.

The engine is designed to handle complex model structures natively, which means three things: teams spend less time restructuring models for performance, AI agents can operate against governed business data, and simulation works against the model's actual formulas, dependencies, and live data. Combined with optimization and statistical forecasting, these tools are available to both users and governed agents.

What we're building next

State-based modeling, customer lifecycle analysis, pipeline conversion modeling, subscription tier migration, AR aging, and revenue forecasting can all work against live model data. The architecture is designed to support capabilities that go beyond traditional planning, including causal analysis and risk-adjusted optimization, so teams can understand not just what happened, but why.

See it in action

The best way to understand the difference is to use it. Build your first model in minutes.

Request Pilot Access Back to Overview

Working with select organisations in pilot