AI with a model of the world underneath it

Harness AI.
Keep reality in the loop.

ONEMODEL is CANCEA’s AI-assisted digital twin and agent-based scenario modelling platform. Move from a question to governed data and executable scenarios that account for people, firms, markets, institutions, infrastructure, and time—then publish results with a traceable evidence chain.

Explore the platform → Request a demo
Model fasterUse AI to accelerate structure and dynamics authoring while keeping human review in the loop.
Simulate interactionsRun scenarios through a digital twin where households, firms, markets, institutions, and constraints interact.
Trace every resultConnect governed inputs, run provenance, scenarios, and published outputs in one evidence chain.
From intent to evidenceGoverned execution
01 · ONEMODEL Studio
Start with the question
Translate expert intent into model structure, decisions, and scenarios through model-aware AI tools.
↓
02 · ONEMODEL Data
Bind exact inputs
ONEMODEL Data provides versioned, queryable source data with explicit time context and provenance.
03 · ONEMODEL Studio
Freeze what will run
The model package, history, inputs, scenarios, and execution identity are explicit and reproducible.
↓
04 · ONEMODEL Engine
Let the system interact
ONEMODEL Engine executes the digital twin so outcomes emerge from interacting agents, institutions, markets, and constraints.
PeopleFamiliesDwellingsBusinessesGovernmentInfrastructureMarkets
Baseline
→
Alternative scenarios
↓
05 · ONEMODEL Results
Publish evidence, not just answers
ONEMODEL Results publishes immutable releases that preserve scenario, trial, model, input, and run provenance.
Governed inputs•Run identity•Scenario lineage•Published release
AI-assisted authoringfrom intent to model structure
Governed inputsversioned data with provenance
Executable digital twininteracting agents and institutions
Scenario intelligencecompare paths, not just point estimates
Traceable publicationimmutable, auditable results
Why ONEMODEL

AI is powerful. Complex decisions still need a world model.

Language models can accelerate synthesis, authoring, and interpretation. They do not replace explicit state, constraints, behavioural logic, or reproducible execution. ONEMODEL combines AI with a digital twin so complex questions can be tested against a coherent model of how the system works.

01

Turn expertise into an executable model.

Use natural language and model-aware AI tools to accelerate the translation from domain knowledge into agents, traits, decisions, and scenarios.

02

Constrain intelligence with reality.

Ground the model in governed data, explicit relationships, initial state, and system constraints before asking it to project forward.

03

Compare consequences, not just answers.

Run alternative scenarios through the same digital twin and follow how effects propagate across people, markets, institutions, places, and time.

Proven in practice

Already supporting decisions across governments, research, and municipalities.

ONEMODEL is being applied to real institutional decisions where departments, funding portfolios, municipal finances, infrastructure, populations, and time need to be considered together.

Used in practiceReal deployments. Real institutional scale.
PEIGovernment of
Prince Edward Island
CIHRCanadian Institutes
of Health Research
AMOAssociation of
Municipalities of Ontario
26PEI government departments connected to a shared planning framework
5 yearsCIHR engagement evaluating impacts associated with funded research
444Ontario municipalities represented in the AMO analysis
Real deployment · Whole-of-government planning↗
26provincial departments

Government of Prince Edward Island

A shared population and economic planning system connecting demographic change, departmental revenues and expenses, investment decisions, public services, infrastructure, and policy scenarios across government.

Explore the PEI deployment
Real deployment · Research-funding evaluation↗
5year engagement

Canadian Institutes of Health Research

ONEMODEL is being used to evaluate the impacts associated with CIHR-funded research, applying agent-based modelling to a complex attribution problem that unfolds across institutions, people, outcomes, and time.

Explore the CIHR deployment
Real deployment · Municipal finance + infrastructure↗
444Ontario municipalities

Association of Municipalities of Ontario

A province-wide model joining full municipal revenue and expenditure accounts with infrastructure value, condition, renewal, growth needs, reserves, debt, taxes, user fees, transfers, and demographic drivers.

Explore the AMO deployment
How it works

From question to published scenario in one governed flow.

AI helps at the points where language and expert intent matter most. Deterministic systems take over where reproducibility, execution, and evidence matter most.

One connected system

A model you can follow end to end.

ONEMODEL connects model authoring, governed data, execution, run identity, and published results so the final answer can be traced back through the system that produced it.

Layer01

Authoring. Define the system, assumptions, decisions, scenarios, and data requirements.

Layer02

Execution. Run the same model identity across history, baseline, and alternative scenarios.

Layer03

Evidence. Preserve provenance from governed inputs through run identity to published results.

01

Author in ONEMODEL Studio

Analysts can move from natural-language intent to draft model structure, and from reviewed dynamics proposals to code generation. AI accelerates authoring, while model definition and approval remain explicit.

Available
02

Bind with ONEMODEL Data

ONEMODEL Data governs input query and validation semantics, including canonical query metadata and explicit as_of context. Inputs are selected deliberately rather than pulled opaquely from an AI black box.

Available
03

Freeze the run in ONEMODEL Engine

ONEMODEL Engine fixes the model package, run plan, history identity, inputs, scenarios, and execution semantics so each run can be reproduced and compared consistently.

Available
04

Execute in ONEMODEL Engine

ONEMODEL Engine runs the model outside the language model. AI can help define the system, but deterministic simulation remains the engine of record.

Available
05

Publish with ONEMODEL Results

Verified run packages carry provenance and fingerprints. ONEMODEL Results turns outputs into immutable published releases while preserving the evidence chain behind them.

Operational
Capabilities

One platform from model design to governed results.

ONEMODEL brings AI-assisted authoring, governed data, executable simulation, scenario management, and results publication into one connected modelling environment.

A
Available

ONEMODEL Studio

Natural-language prompts can be translated into model structure through bounded, model-aware tools. Expert teams move faster while retaining explicit review and control.

B
Available

AI-assisted dynamics

Within ONEMODEL Studio, decision and enactment logic can be proposed in a staged review flow. Human approval remains central, and validation is part of the authoring workflow.

C
Available

ONEMODEL Data

Governed source data, canonical query contracts, explicit time context, and provenance keep every model tied to explicit, versioned inputs.

D
Available

ONEMODEL Engine

Deterministic execution, model and input fingerprints, run identity, and compatibility checks make scenarios reproducible and comparable.

E
Available

ONEMODEL Results

A governed output layer publishes immutable releases while preserving scenario, trial, model, input, and run provenance for downstream analysis and review.

F
Available

Full-system digital twin

Represent interacting populations, households, businesses, governments, infrastructure, and markets in a common stateful simulation rather than isolating them in separate forecasts.

The digital twin

Canada as a living system, not a spreadsheet.

The flagship ONEMODEL national digital twin connects people, families, dwellings, businesses, governments, infrastructure, and markets inside one evolving population-economic system. Scenarios change the rules or conditions; the model carries those changes through the interacting system over time.

ONEMODELinteracting agents and institutions
Peoplelife-course dynamics
Familieshousehold structure
Dwellingshousing supply and use
Businessesproduction, labour, decisions
Governmenttaxes, transfers, programs
Infrastructurecapacity and constraints
Marketshousing · labour · goods

See the interactions that averages hide.

When a change affects households, firms, governments, capacity, prices, labour, or location at the same time, a collection of disconnected forecasts can miss how effects compound. A digital twin keeps those relationships inside the same evolving system.

  • Follow distributional effects. Examine how outcomes differ across people, households, industries, or places.
  • Capture feedback. Let changes in one part of the system alter conditions elsewhere.
  • Compare paths through time. Test not only where a scenario ends, but how the system gets there.
  • Keep the evidence chain visible. Tie results back to data, model identity, scenarios, and run provenance.
Model applications

From proven deployments to new complex systems.

The real deployments above demonstrate the platform in practice. The ONEMODEL Platform can also represent very different systems. ONEMODEL began as CANCEA’s integrated national model; the surrounding data, authoring, execution, and publication capabilities evolved into the ONEMODEL Platform. Energy, health-system, and supply-chain examples illustrate how that platform generalizes when interactions, feedback, constraints, and time materially change the answer.

Flagship national model

ONEMODEL

Economy, households, housing, infrastructure, social value, and well-being.

Current

Connect policy and investment choices to economic, fiscal, distributional, health, and quality-of-life outcomes inside one evolving population-economic system.

Explore ONEMODEL
Illustrative application

Energy Transition Model

Generation, storage, grids, utilities, demand, investment, and policy.

Test how demand growth, technology choices, transmission constraints, prices, reliability, and emissions interact across an evolving energy system.

Explore the energy model
Illustrative application

Health System Model

Population health, care delivery, workforce, capacity, costs, and well-being.

Follow how disease, access, treatment, workforce capacity, employment, household finances, and quality of life affect one another over time.

Explore the health model
Illustrative application

Supply Chain Resilience Model

Suppliers, inventories, logistics, infrastructure, labour, and recovery.

Trace disruptions through sourcing, inventory depletion, production, substitution, prices, employment, and recovery paths.

Explore the supply-chain model
Where it fits

Not every question needs a digital twin.

ONEMODEL is built for decisions where interactions, constraints, behaviour, and time materially change the answer. When the problem is simpler, the tool should be simpler too.

Use ONEMODEL when the system matters

  • Decisions with direct, indirect, induced, and feedback effects.
  • Distributional questions across people, households, industries, or places.
  • Scenarios where the path matters, not just the final total.
  • Problems where behaviour, capacity, affordability, and institutions interact over time.
  • Situations where decision-makers need a traceable line from data to result.

Use a simpler tool when the system does not

  • Static dashboards or descriptive reporting.
  • Simple forecasts with weak system feedback.
  • Cases where available evidence does not justify model resolution.
  • Very fast decisions where calibration time would dominate value.
  • Problems where clarity of assumptions matters more than structural richness.
“AI proposes. Humans approve. Data grounds. The digital twin executes.”ONEMODEL design principle
Built for connected questions

When one decision propagates across a system.

ONEMODEL is designed for questions that cross institutional and sector boundaries—where outcomes depend on who is affected, how behaviour changes, and which constraints bind over time.

People & communitiesPopulation, households, affordability, well-being, and distributional outcomes.
Economy & workforceEmployment, business activity, labour markets, incomes, and fiscal effects.
Scenario intelligence

Test the path before committing to it.

Compare baseline and alternative scenarios under the same model identity, with governed inputs and a reproducible execution path.

Infrastructure & capacityInvestment, timing, constraints, service needs, and downstream economic effects.
Policy & programsBehavioural response, eligibility, fiscal flows, and direct and indirect consequences.
ONEMODEL FAQ

What ONEMODEL is—and what it is not.

Plain-language answers to the distinctions that matter when evaluating the platform.

What is ONEMODEL?

ONEMODEL is CANCEA’s model-agnostic platform for AI-assisted model authoring, governed data, deterministic digital-twin and agent-based simulation, scenario management, provenance, and traceable results publication.

Does AI generate the simulation results?

No. AI can assist with model design, authoring, and interpretation, but executable model logic and governed inputs run through the deterministic simulation engine. Human review and validation remain part of the modelling workflow.

Is the ONEMODEL Platform the same as the national digital twin?

No. The ONEMODEL Platform is model-agnostic infrastructure. The ONEMODEL National Digital Twin is the flagship socioeconomic model built on that platform and is one example of what the platform can support.

When is a digital twin useful?

It is most useful when interactions, feedback, constraints, distributions, geography, institutions, or time materially change the answer. For simpler descriptive or forecasting questions, a simpler analytical tool may be more appropriate.

See ONEMODEL in action

Ask a complex question.
Run it through reality.

See how ONEMODEL moves from model intent to governed data, executable scenarios, and traceable published results—without asking AI to stand in for the model itself.

Request a demo → Explore the workflow