AI Decision Intelligence for Government

Ask your city.
Test decisions before you spend.

TwinGov turns government documents, GIS, public data and operational information into a source-grounded digital model you can question, explore and use to test decisions.

  • Answers cite the source they came from
  • Starts from the data you already hold
  • Missing data is reported, not filled in
TwinGov / Community model
Concept
01 kmN

Ask

Where should we locate a new community facility to maximise 15 minute access?

Residents reached

+18,400

Coverage gain

+11.2%

Confidence

Moderate

  • Candidate site
  • Existing facilities
  • 15 min walk

Product interface concept. Geometry and figures are illustrative and describe no real jurisdiction.

The environment

Map, model and evidence in one view.

The question, the geography and the sources behind the answer stay on the same screen. Nothing has to be reassembled by hand afterwards.

TwinGov / Scenario Lab
Community modelConcept
01 kmN
  • Candidate site
  • 15 min walking access
  • Existing facilities
  • Area with a funded project

Question

Where should we locate a new community facility to maximise 15 minute access?

Result

One candidate area outperforms the other three on residents reached within a 15 minute walk. The advantage comes from the street network, not from raw distance: two of the higher scoring sites on a straight-line measure are separated from their catchment by the rail corridor.

Supporting metrics

Residents reached

18,400

Within a 15 minute walk

Derived

Current service gap

11.2%

Population outside any catchment

Derived

Walking accessibility

12.4min

Median from the candidate site

Derived

Nearby infrastructure

4assets

Transit, utility and road capacity

Measured

Sources

  • Transportation networkGIS layerPeriodicMeasured
  • Population datasetDatasetCensus cycleMeasured
  • Existing facilities inventoryDatasetMaintained by the institutionMeasured
  • Local development planDocumentOn adoptionMeasured

Declared gap in this answer

Daytime population is not connected, so the estimate covers residents only. Workers and visitors are outside the model.

Product interface concept. The map geometry is invented and the figures are illustrative. TwinGov displays no data about a jurisdiction until that jurisdiction connects its own sources.

What TwinGov does

One model. Every decision starts with better context.

Five capabilities over a single model of your community. Not five tools that have to be reconciled afterwards.

  • 01

    City Intelligence

    Ask questions across documents, datasets, indicators and maps at once.

    A plain-language question is answered from the model itself: the strategy document, the indicator series and the map layer are read together, and the answer states which of them it relies on.

    Question to evidence in one step

  • 02

    Geospatial Intelligence

    Understand where problems, services, infrastructure and opportunities actually are.

    Thematic views over mobility, climate, safety, energy, heritage and land use, plus coverage analysis that shows which recognised problems have a funded project behind them and which do not.

    Walking access on the real street network

  • 03

    Scenario Lab

    Test proposed investments and interventions before committing resources.

    Place a candidate facility on the map and see the population it reaches, computed as an isochrone over the street network rather than a circle drawn around a point. Save it, build an alternative, compare them.

    Compare alternatives side by side

  • 04

    Decision Reports

    Turn an analysis into concise, source-backed material for a meeting.

    A report carries the map, the figures, the assumptions and the source list, in a form that can go into a committee file or a funding application without being rebuilt by hand.

    Export as document or PDF

  • 05

    Government Data Model

    Bring fragmented institutional information into one structured model.

    Areas, services, infrastructure, projects, indicators and recognised problems become a schema-validated model, where every element carries its source, its refresh frequency and its confidence.

    Open JSON schema, GeoJSON layers

From data to decision

Five steps, and the evidence survives all five.

Nothing in this sequence asks a government to rebuild its information infrastructure first.

  1. 01

    Connect

    Government documents, GIS layers, open data and operational datasets.

    Strategic and comprehensive plans, zoning and land use, capital programmes, budgets, demographic series, infrastructure inventories, environmental readings. Whatever exists is connected where it already lives.

  2. 02

    Model

    TwinGov structures areas, services, infrastructure, projects, indicators, problems and the relationships between them.

    Each element records where it came from and how current it is. Coverage is measured, and what the model does not know is written down as a declared gap rather than left implicit.

  3. 03

    Ask

    Users ask questions in natural language.

    The answer is assembled from the connected model, cites the documents, layers and indicators behind it, separates measured fact from assumption, and says plainly when the data cannot support a conclusion.

  4. 04

    Simulate

    Test interventions geographically and compare the alternatives.

    Position a proposed investment, measure who it reaches over the real network, then set it against a second option on the same indicators.

  5. 05

    Decide

    Export evidence-backed findings for meetings, planning and investment decisions.

    The reasoning leaves the platform intact: the figures, the map, the assumptions and the sources travel together, so the decision can be explained to anyone who asks later.

Source-grounded AI

AI that shows its work.

A generative model can produce a confident answer from thin evidence. That is the failure mode a public institution cannot afford, so TwinGov is built the other way round: sources first, conclusions second.

  • Every claim points back

    An important statement can be traced to the document, dataset, map layer or indicator that supports it. The citation is part of the answer, not a footnote added afterwards.

  • Fact stays separate from estimate

    Measured values, derived values and assumptions are labelled differently and stay distinguishable all the way into the exported report.

  • Missing data is a finding

    When the connected sources cannot answer a question, TwinGov says so and names what is missing. A declared gap is more useful than a plausible number.

  • The decision stays with the person

    TwinGov is decision support. It does not approve investments, and it does not replace planners, engineers or administrators who are accountable for the outcome.

City Intelligence
Grounded answer

Question

Which areas should be prioritised for additional green infrastructure?

Answer

Three districts combine above-average heat exposure with the lowest green space per resident in the model. The northern district is the strongest candidate on both measures at once; the two others rank high on one measure each, which is a different case to argue.

Evidence

  • Heat exposure in the three districts sits in the top quartile of the connected series.
  • Green space per resident in the same districts is below the model median.
  • Two of the three already appear as recognised problems in the adopted strategy, without a funded project attached.

Districts flagged

3

Derived

Residents in scope

42,100

Derived

Green space

6.8m² / res

Measured

Sources used in this answer

  • Land surface temperature seriesDatasetDaily, history retainedMeasured
  • Green space and land coverGIS layerPeriodicMeasured
  • Population by districtDatasetCensus cycleMeasured
  • Adopted strategy, environment chapterDocumentOn adoptionMeasuredTwo of the three districts are named in the recognised problems list.

ConfidenceModerate

Heat and land cover are well covered. The ranking between the second and third district rests on a single indicator and should not be treated as settled.

What the data does not cover

  • Tree canopy inventory is not connected, so shading is inferred from land cover rather than measured.
  • Irrigation capacity and maintenance cost are outside the model and would change the feasibility of any of the three.

Product interface concept. Figures are illustrative and describe no real jurisdiction.

Scenario Lab

Test the decision before funding it.

Place a proposed intervention on the map, measure who it actually reaches, then put it next to the alternative. The comparison runs on the same population, accessibility and infrastructure data for both options.

What can be placed

  • Schools and early childhood
  • Community facilities
  • Transit stops
  • Parks and green space
  • EV charging
  • Cooling centres
  • Primary healthcare access
  • Emergency services
  • Mobility infrastructure
  • Utility and public works

What the comparison reports

  • Population served
  • Geographic coverage
  • Walking accessibility
  • Existing service gaps
  • Vulnerable populations
  • Infrastructure proximity

Access is computed as an isochrone over the real street network. A circle drawn around a point flatters any site that sits next to a river, a rail line or a motorway, and those are exactly the sites that look best on a map and disappoint after they are built.

See how Scenario Lab works
Scenario Lab / Comparison
Same inputs, both options
AScenario A

Community facility on the western district edge, adjacent to an existing transit stop.

BScenario B

Community facility in the central district, closer to the population centroid.

  • Population served, 15 min walk

    18,400
    14,900
  • Geographic coverage added

    11.2%
    7.4%
  • Median walking time

    12.4min
    9.8min
  • Residents currently unserved

    9,100
    3,600
  • Transit stops within catchment

    6
    11
  • Overlap with existing catchments

    14%
    38%

Difference

A reaches 3,500 more residents and 2.5 times as many people currently outside any catchment. B is faster to reach for those it serves and sits on a denser transit node, but most of its catchment is already covered by an existing facility.

TwinGov reports the difference. The choice, and the responsibility for it, stay with the institution.

Product interface concept. Figures are illustrative and describe no real jurisdiction.

Questions it is built for

The questions that hold up a decision.

These are the shapes of question TwinGov is designed to answer against a connected model of your own community.

  • 01

    Where should we locate the next public facility to serve the greatest number of residents?

  • 02

    Which neighbourhoods have the largest infrastructure gaps?

  • 03

    Which areas are underserved by public transportation?

  • 04

    What communities are more than a 15 minute walk from essential services?

  • 05

    Which proposed capital investment creates the highest coverage improvement?

  • 06

    What areas have the greatest heat exposure and the lowest access to green space?

  • 07

    How does Scenario A compare with Scenario B?

  • 08

    Which projects in our capital plan address the priorities in our strategic plan?

  • 09

    What data supports this recommendation?

  • 10

    What information is missing before this decision can be made confidently?

These are question shapes, not stored answers. What TwinGov can say about any one community depends entirely on the sources that community connects, and the platform reports the difference rather than filling it in.

Use cases by department

The same model, read from where you sit.

  • Prioritise investments and prepare evidence for decisions.

    • Compare candidate investments on identical indicators
    • See which recognised problems still have no project behind them
    • Take a source-backed brief into the meeting
    A question from this desk

    Which of the four proposals in front of the council reaches the most residents for the money?

Data ownership and interoperability

Your data stays your data.

A decision-intelligence layer is worth very little if it becomes the only place the information can be read. TwinGov is built so the institution can walk away with the model it helped build.

  • The institution keeps control

    Data supplied by the institution remains under its control, on terms set in the agreement rather than by the platform.

  • Open formats, not a proprietary container

    The model exports as JSON against a public schema. Geographic layers export as standard GeoJSON, readable in any GIS.

  • It sits above your GIS, not in front of it

    Existing geographic services are connected as sources. A WFS service can expose model layers back to the tools your teams already use.

  • API-ready by design

    The same structured model that the interface reads is available programmatically, so the institution is not dependent on the interface to get value out.

Third-party layers under their own licences, such as commercial traffic feeds, satellite imagery or terrain tiles, are not transferable with the model. Layers derived from OpenStreetMap carry the ODbL licence and its attribution.

Export
Open formats
  • community-model.jsonFull model against a public schemaJSON
  • areas.geojsonTerritorial units and boundariesGeoJSON
  • services.geojsonFacilities, catchments and coverageGeoJSON
  • indicators.jsonSeries with source and refresh frequencyJSON
  • scenarios.jsonSaved simulations with their assumptionsJSON
  • WFS endpointModel layers straight into your GISOGC

No conversion step, no proprietary container, and no need to ask the vendor for a copy of your own model.

Product interface concept. Export scope is set in the agreement and respects the licences of the underlying sources.

Built around existing government data

Start with the data you already have.

Deployment does not begin with a data programme. It begins with the documents and datasets an administration already produces, and the model gets more useful as more of them are connected.

  • Strategic plansDocument
  • Comprehensive plansDocument
  • Zoning and land useDocument + GIS
  • Capital improvement plansDocument
  • BudgetsTabular
  • GIS layersGeospatial
  • Demographic datasetsTabular
  • Infrastructure inventoriesTabular + GIS
  • Mobility dataGeospatial
  • Environmental dataTime series
  • Open data portalsAPI
  • Institutional datasetsMixed

No sensor network is required to start, and no building model. A first version of the model is built from what an administration already publishes and already maintains. It becomes more useful as further institutional data is connected, and the interface shows how much of the model each addition fills.

Government data model
8 of 12 connected

Model coverage

68%

How much of the model the connected sources fill. It does not measure how accurate or how current any single value is.

  • Comprehensive planDOC
  • Zoning and land useGIS
  • Capital improvement planDOC
  • Street networkGIS
  • Population by areaSET
  • Facilities inventorySET
  • Transit routes and stopsGIS
  • Air quality seriesAPI
  • Utility networkGIS
  • Pavement conditionSET
  • Daytime populationSET
  • Tree canopy inventoryGIS

Declared gaps

  • Daytime population is absent, so access analysis covers residents only.
  • Utility network is connected for two of five districts.
  • Pavement condition predates the last resurfacing programme.

Product interface concept. The source list and coverage figure are illustrative.

Who uses TwinGov

Six seats around the same model.

  • City / County Executive

    Understand priorities and compare investments.

    Sees what the institution has recognised as a problem, what is funded against it, and how two proposals differ on the indicators that will be asked about in public.

  • Planning Director

    Evaluate locations and spatial impact.

    Tests a site against population, access and land use before the location is committed, with the plan documents read alongside the map.

  • GIS Manager

    Connect existing geographic information to a broader decision model.

    Keeps the authoritative layers where they are. TwinGov consumes them, returns standard formats, and does not become a second copy to maintain.

  • Public Works Director

    Identify service and infrastructure gaps.

    Reads coverage and condition together, and can show where a gap is measured rather than reported.

  • CIO / CTO

    Create a usable intelligence layer over fragmented institutional data.

    One structured model over systems that do not talk to each other, with roles, auditability and an exit path in open formats.

  • Analyst

    Research questions across documents, maps and datasets without searching five systems by hand.

    Gets to the evidence quickly, and gets the citation with it, so the finding can be checked by someone else.

Trust

Built for decisions that need to be explained.

Public decisions get questioned later, sometimes years later, sometimes in front of an auditor. The platform is designed for that moment.

  • Source traceability

    Every significant claim resolves to the document, dataset or layer behind it.

  • Data ownership

    The institution keeps control of what it supplies and can export the model.

  • Role-based access

    Users see what their role allows, within their institution's workspace.

  • Auditability

    Analyses, scenarios and reports keep the inputs and assumptions they were produced from.

  • Open data formats

    JSON against a public schema, GeoJSON for geography.

  • Human in the loop

    The platform supports the reasoning. A person makes and signs the decision.

  • Transparent limitations

    Coverage, refresh frequency and declared gaps are visible in the interface, not buried.

  • Secure architecture

    Institutional workspaces are separated, private and not publicly indexed.

TwinGov holds no security or compliance certification at this time, and none is claimed on this site. Where a procurement process requires formal attestation, say so early and we will tell you plainly what exists and what does not.

Read the trust overview

Demonstration

Explore a live city model.

See how TwinGov connects geography, public data and decision intelligence in one environment.

  • The model, the map and the sources in one environment
  • A worked question with its evidence and its declared gaps
  • A decision report exactly as it leaves the platform
Coverage view
Problems against projects
01 kmN
  • Documented project in place
  • Recognised problem, no funded project
  • Existing facilities

The view measures whether a project exists and how far it has got, not whether the institution is right to have prioritised it.

Decision report
PDFDOCX
TwinGov
Decision report

Community facility siting

Comparison of four candidate locations on population reached, coverage added and existing service gaps.

  1. 1The question and how it was framed
  2. 2Candidate locationsFour sites, same method
  3. 3Population reached and coverage added
  4. 4Assumptions used
  5. 5Sources
  6. 6What this analysis does not coverDeclared before conclusions, not after
7 sources cited2 declared data gaps

Product interface concept. Geometry and figures are illustrative and describe no real jurisdiction.

What could TwinGov reveal about your community?

Start with the data you already have. We can build an initial model from available public and institutional sources and show where TwinGov can create immediate decision value.