Methodology

How Landvex works.

Five interlocked layers — from global field observation to verified intelligence. Every output is sourced, explained, and challenged before delivery.

Request a briefing → Download methodology PDF →
01

Observation
Layer

02

Normalization
Layer

03

Intelligence
Layer

04

Contradiction
Engine

05

Red Team
Analysis

Verified
Output

Landvex Intelligence Methodology

8-page technical overview. How field observations become decision intelligence.

Download PDF
1
Observation Layer

Eyes on the ground.
At global scale.

Every Landvex intelligence product begins with physical observation. The quiXzoom contributor network gives us human eyes across more than 20 countries — not cameras, not satellites, but verified people on the ground documenting what is actually there.

🌐

quiXzoom global contributor network

Thousands of active field contributors across 20+ countries, deployable within hours to any target area. Local knowledge combined with structured data collection.

🤖

AI-reviewed submissions

Every submission is automatically reviewed for geographic accuracy, image quality, and specification compliance before it enters the pipeline. No human bias in the review gate.

🪪

Verified field agents, identity-verified once

Every contributor completes a one-time identity verification. Verified status persists across all missions. Verification creates accountability and trace integrity for every observation.

🗺️

Coverage: 20+ countries at launch

Europe, Southeast Asia, and growing. Coverage expands continuously as the quiXzoom network scales. New markets are opened on request.

2
Normalization Layer

Raw observation becomes
structured data.

Field observations arrive in different formats, from different contributors, across different contexts. The normalization layer converts everything into a consistent, comparable structured data point — stripping noise while preserving source integrity.

📐

Every observation converted to a structured data point

Unstructured field submissions are parsed into typed, schema-validated records. Location, object, condition, and change signal are extracted and stored independently.

🔍

Entity extraction: location, object type, condition, change signal

Each data point is tagged with what was observed, where, in what condition, and whether it represents a change from the prior state — the four axes of physical intelligence.

🕐

Temporal indexing: every observation timestamped and versioned

Every data point is stored with a precise timestamp and version history. This enables change detection, trend analysis, and before/after comparisons at any granularity.

🔗

Source attribution maintained throughout

Every data point retains a full audit trail back to the field submission, the contributor, the timestamp, and the mission brief. Nothing is anonymized into oblivion.

Every place scored on six axes.

Observations aggregate into district, city, and regional scores across six dimensions — all on a 0–100 scale, all updated continuously.

💡
Opportunity
0–100
📈
Growth
0–100
🏪
Commercial Vitality
0–100
🔩
Infrastructure Stability
0–100
✈️
Tourism Potential
0–100
💎
Investment Confidence
0–100
3
Intelligence Layer

Observations aggregate
into scores.

Individual data points are aggregated into district, city, and regional intelligence scores. Scores are updated continuously as new observations arrive — not quarterly, not annually. The physical world changes in real time; the intelligence layer follows it.

🏙️

Observations aggregated into district / city / region scores

Multiple observations from multiple contributors covering the same area are weighted, de-duplicated, and aggregated into a single composite score per dimension per area.

🔄

Scores updated continuously as new observations arrive

There is no batch cadence. Every validated observation triggers a score recalculation for the relevant area. Intelligence is always as fresh as the most recent field data.

📊

Six dimensions, one comparable scale

Opportunity · Growth · Commercial Vitality · Infrastructure Stability · Tourism Potential · Investment Confidence. Bangkok vs Kuala Lumpur vs Milan — on the same axes, the same scale.

4
Contradiction Engine

Official data vs.
observed reality.

The most valuable intelligence is not confirmation — it is contradiction. The Contradiction Engine ingests official data sources and actively searches for conflicts with what our contributors actually observed. When the data says one thing and the ground says another, that is where the intelligence lives.

🏛️

Official data ingested continuously

National statistics, municipal open data, land registries, and transport authorities are continuously ingested and indexed against covered geographies.

⚖️

Every official claim compared against observed reality

For every official data point covering an area with observation coverage, the Contradiction Engine runs an automated comparison. Supporting evidence and contradicting evidence are catalogued separately.

🚩

Contradictions flagged, ranked by severity and economic impact

Not all contradictions are equal. The engine ranks flagged conflicts by how severe the discrepancy is, and by the estimated economic significance of the affected area or asset class.

📐

Confidence scores from evidence weight

Every contradiction report includes a confidence score calculated from the volume, recency, and source diversity of the supporting and contradicting evidence.

5
Red Team Analysis

Every conclusion
challenged before delivery.

Before any major conclusion reaches a client, a second AI model actively attempts to disprove it. This is not a review — it is an adversarial process. If the conclusion survives, it is delivered. If it doesn't, it is revised or flagged.

⚔️

Every major conclusion challenged by a second model

The primary analysis model produces a conclusion. A separate red team model, given the same evidence, is tasked with finding holes, alternative explanations, and disconfirming observations. The conclusion is only finalized after this adversarial review.

Supporting evidence listed

Every delivered conclusion includes a full list of the observations, data points, and sources that support it — with source attribution and timestamps. Clients see what the conclusion is based on.

Contradicting evidence listed

Contradicting evidence is not suppressed. Every conclusion is delivered alongside the evidence that pushed against it. Where the red team found weaknesses, those are explicitly documented.

📏

Uncertainty quantified

Confidence levels are expressed numerically, not as qualitative hedges. Clients know not just what Landvex concludes, but how certain that conclusion is and why.

What Landvex delivers.

Six output types. All sourced. All explained. All ready for decisions that cost money.

📊

Scores (0–100 per dimension)

Composite intelligence scores per location, across six dimensions. Updated continuously. Comparable across geographies and over time.

Every score cites contributing observations and source data.
🚩

Contradiction Reports

Where observed reality conflicts with official data or investment thesis. Ranked by severity and economic significance. Evidence on both sides included.

Includes confidence score and source breakdown.
💡

Opportunity Reports

Where physical conditions signal under-exploited potential. Targeted by geography, asset class, or decision type. Ranked by opportunity magnitude.

Field observations and trend data cited throughout.
⚠️

Risk Forecasts

Where physical conditions signal emerging risk — before it appears in official statistics or financial reporting. Predictive scoring based on observed change velocity.

Uncertainty ranges included with every forecast.
📝

Executive Summaries

Decision-ready narratives structured around the client's specific question — not generic reports. Concise, direct, and directly linked to the decision being made.

Supporting data attached; key sources highlighted.
📌

All outputs cite sources
and explain reasoning

There is no black box. Every Landvex output tells you what it concluded, what it was based on, what pushed against it, and how confident the conclusion is.

Because intelligence you can't verify is just opinion.

The standards behind every score.

Intelligence is only as good as the data it is built from. Every observation in the Landvex pipeline is subject to explicit accuracy standards, a documented verification chain, and a confidence scoring system.

Accuracy standards

Every field observation is held to explicit accuracy requirements before it is admitted into the pipeline. Rejection rates are tracked and published internally as quality metrics.

  • Geo-location verified against reported coordinates
  • Photograph metadata validated (device, timestamp, GPS)
  • Submission reviewed by AI against mission brief
  • Duplicate detection across concurrent contributors
  • Outlier flagging when observation deviates significantly from area baseline

Verification chain

Every observation has a complete audit trail. From contributor identity through submission review to final score contribution, every step is logged and immutable.

  • Contributor identity verified once, applied to all submissions
  • Submission hash logged at ingestion (tamper detection)
  • Review decision and reviewer logged with timestamp
  • Score contribution traceable back to source observations
  • Contradiction reports include full evidence chain on both sides

Confidence scoring

Every score, conclusion, and contradiction report carries a numerical confidence value. Confidence is calculated from evidence volume, recency, and source diversity — not subjective assessment.

  • Volume factor: more observations → higher confidence
  • Recency factor: older data decays on configurable half-life
  • Diversity factor: independent sources weighted above correlated ones
  • Contradiction penalty: conflicting evidence reduces confidence proportionally
  • All confidence values expressed numerically (0–100)

How we detect contradictions.

A six-step pipeline that runs continuously against every area with active observation coverage. When official data and observed reality disagree, this is what happens.

1. New field observations ingested

Validated submissions enter the pipeline, tagged with location, object type, condition, and change signal

2. Official data sources queried for same geography

National statistics, municipal open data, land registries, and permit databases are checked for the affected area

3. Contradiction detection: observed vs. official

AI model compares observed conditions against official claims on matching dimensions (vacancy, activity, condition, growth)

No contradiction

Evidence supports official data → confidence score updated upward

Contradiction detected

Divergence ranked by severity and economic significance

5. Red Team challenge

Second AI model attempts to disprove the contradiction with alternative explanations. Conclusion revised or confirmed.

6. Contradiction Report delivered

Client receives: finding, supporting evidence, contradicting evidence, severity rank, confidence score (0–100)

Ready to see it applied
to your decision?

Tell us the question you need answered. We'll show you how Landvex can support it.

Request a briefing →