AI Capability Platform — the engine layer that transforms raw observations into structured intelligence through specialised AI engines.

AAMOS (Adaptive Autonomous Multi-Observation System) is not a single AI system — it is a platform of composable AI engines. Each engine handles a specific domain: Vision for object detection, Identity for verification, Fraud for anomaly detection, Infrastructure for asset condition, and more. Developers access AAMOS through REST APIs that accept images, video, or structured data and return verified intelligence. AAMOS is developed and sold by Landvex Inc, Houston, TX — there is no "AAMOS Group."

The end-to-end process for training, validating, and deploying computer vision models across AAMOS engines.

The AAMOS training pipeline ingests labelled datasets (from SceneNet, ContribNet, and verified field observations), trains models using YOLO and other architectures, validates against hold-out test sets, and exports production-ready ONNX models. The pipeline includes automated data augmentation, hyperparameter tuning, and model drift detection. Every deployed model carries a version tag and confidence calibration curve.

Application Programming Interface — the programmatic interface through which developers access AAMOS engine capabilities.

The AAMOS API is a REST-style HTTP API that exposes all engine capabilities as standardised endpoints. Authentication uses OAuth 2.0. Responses are JSON with structured confidence scores, bounding boxes, and provenance metadata. Webhooks deliver asynchronous results for batch operations. Rate limits and usage quotas are enforced per API key.

AI engine for regulatory compliance checks, documentation verification, and standard adherence detection.

AMOS Compliance analyses visual and document evidence against regulatory frameworks: safety signage requirements, accessibility standards, environmental permits, and building codes. It identifies missing elements, non-conforming installations, and documentation gaps. Outputs include compliance scores, violation lists with regulatory references, and remediation priorities.

AI engine for detecting visual fraud: skimming devices, manipulated meters, forged documents, and insurance scams.

AMOS Fraud applies computer vision and document analysis to detect tampering, overlays, and anomalous modifications in physical assets and documents. It identifies ATM skimmers by comparing against reference images, detects manipulated utility meters through seal verification, and flags forged identity documents via microprint and font analysis. Each finding includes a fraud probability score and recommended action.

AI engine for identity verification, document authentication, liveness detection, and facial comparison.

AMOS Identity verifies that a person is who they claim to be. It analyses government-issued IDs for forgery indicators, performs liveness checks to prevent spoofing (photos, masks, replay attacks), and compares facial biometrics against the ID photo. The engine outputs a verification confidence score, spoof probability, and document authenticity rating. It powers KYC flows for quiXzoom and enterprise customer onboarding.

AI engine for analysing roads, bridges, buildings, poles, signs, and other physical infrastructure assets.

AMOS Infrastructure ingests visual observations of built assets and outputs condition assessments: crack detection, corrosion scoring, vegetation encroachment, signage legibility, and structural anomaly flags. It compares current observations against historical baselines to detect change and deterioration trends. Outputs include per-asset condition indices, maintenance priority rankings, and georeferenced anomaly maps.

General-purpose AI engine for quality control, industrial inspection, and defect detection across manufacturing and logistics.

AMOS Inspection applies computer vision to quality assurance scenarios: surface defect detection on production lines, packaging integrity verification, assembly completeness checks, and logistics damage assessment. Unlike Infrastructure (which focuses on civil assets), Inspection handles manufactured goods, equipment, and industrial processes. It integrates with conveyor systems, drone feeds, and handheld inspection devices.

AI engine for detecting PPE compliance, workplace hazards, and safety protocol violations.

AMOS Safety analyses visual feeds from construction sites, factories, and field operations to identify safety risks: missing hard hats, unguarded machinery, blocked emergency exits, improper ladder use, and chemical storage violations. It generates real-time alerts for immediate intervention and compiles safety scorecards for management review. The engine learns site-specific safety protocols and adapts detection rules per environment.

Core AI engine for object recognition, anomaly detection, and damage classification in visual data.

AMOS Vision is the foundational computer vision engine powering all other AAMOS modules. It performs object detection (YOLO-based), instance segmentation, image classification, and anomaly detection on visual inputs. The engine runs on ONNX-exported models for cross-platform deployment (cloud, edge, mobile). It is the first processing layer for every visual observation entering the AAMOS pipeline.

API endpoint that verifies identity by comparing a live capture against a trusted reference (ID photo, enrolled biometric).

The /authenticate endpoint accepts a live image or video stream and a reference image (e.g., passport photo, previously enrolled face), then returns a match confidence score, liveness verification result, and spoof detection flag. It is the core of identity verification flows and access control systems. The endpoint supports 1:1 comparison only; for 1:N identification, use /identify.

AMOS Vision Orchestrator — real-time guidance system that directs field capture to ensure sufficient evidence quality.

AVO is an orchestration layer that guides field data collection through two phases: Phase A (live guidance) provides real-time feedback to the capture device — "move closer," "adjust angle," "increase lighting" — ensuring the captured media meets quality thresholds before submission. Phase B (post-capture) validates that the submitted evidence satisfies the verified_claim requirements for the specific knowledge pack (vehicle_wheel, building_facade, passport, etc.). AVO is the bridge between raw capture and verified evidence.

A grouped set of missions in the same geographic area, designed for efficient multi-stop field collection.

Batch missions aggregate individual data collection tasks into optimised routes. A Zoomer can complete 8–12 observations in a single 45-minute round, dramatically increasing earnings per hour and reducing travel overhead. Batch missions are the primary mechanism for building corridor density in the quiXzoom network. They appear as a single claimable unit with a unified time window and payout.

AMOS engine that detects differences between current observations and historical baseline data — surfacing what has changed in the physical world.

The Change Engine compares new observations against prior data to identify additions, removals, or modifications to physical assets. It operates at multiple scales: pixel-level differences in repeated photographs, structural changes through computer vision, and semantic changes (e.g., "this was a parking lot, now it's construction"). Each detected change is scored by confidence and severity, and flagged as a Contradiction when it represents an unexpected deviation. The Change Engine powers infrastructure monitoring, property condition tracking, and environmental surveillance.

API endpoint that categorises an image or object into predefined classes with confidence scores.

The /classify endpoint assigns input media to one or more categories from a trained taxonomy. Unlike /detect (which finds objects), /classify answers "what is this?" for the entire input. It supports multi-label classification (an image can belong to multiple classes) and hierarchical taxonomies (e.g., Vehicle → Car → Sedan). Confidence scores are calibrated and include an "unknown" class for inputs that don't match any trained category.

API endpoint that analyses two images of the same asset and reports differences, similarities, and change severity.

The /compare endpoint is the programmatic interface to the Change Engine. It accepts a baseline image and a current image (plus optional GPS and timestamp metadata), aligns them geometrically, and outputs a structured change report: added elements, removed elements, modified regions, severity scores per change, and an overall change index. It is used for automated condition monitoring, insurance claims validation, and regulatory compliance verification.

Descriptive intelligence that tells you the current state of your physical assets — what is, not what to do.

Control Intelligence answers the question: "What is the current condition of my assets?" It is the foundational layer of insight derived from observations, describing reality as it exists today: which bridges are sound, which buildings have damage, which roads need repair. Control Intelligence is objective, measurable, and verifiable. It is the raw material from which Decision Intelligence is forged — you cannot decide what to do until you know what is. Landvex delivers Control Intelligence as indices, scores, and contradiction flags.

A geographic visualisation showing where quiXzoom observations have been collected, updated, and verified.

The Coverage Map displays observation density, freshness, and quality across geographic regions. It answers the critical question for PBOI customers: "How complete is my data?" Coverage is expressed as a percentage of target assets observed within a defined polygon, with colour-coding by refresh SLA compliance. Green = observed within SLA, Yellow = approaching expiry, Red = stale or missing. Coverage Maps are a core component of Landvex subscription offerings.

Actionable recommendations derived from observations through AMOS — the end product of the Landvex ecosystem.

Decision Intelligence is not data. It is not reports. It is the answer to a specific question: Which asset should I repair first? Where should I open my next store? Which claims are fraudulent? Landvex delivers Decision Intelligence by combining quiXzoom observations, AMOS analysis, and domain expertise into structured recommendations that directly enable decisions. Every output includes a confidence score and traceable source data. Decision Intelligence transforms Control Intelligence (what is) into prescriptive guidance.

API endpoint that finds and locates objects within an image, returning bounding boxes, labels, and confidence scores.

The /detect endpoint is the primary object detection interface. It accepts an image and a model identifier, runs inference using the specified YOLO/ONNX model, and returns a list of detected objects with normalised bounding box coordinates (x, y, width, height), class labels, and confidence scores. It supports batch processing for high-throughput scenarios and can be configured with confidence thresholds and NMS (non-maximum suppression) parameters.

AMOS engine that builds traceable evidence chains for audit, insurance, and legal proceedings.

The Evidence Engine constructs immutable, cryptographically signed chains of custody for every observation and analysis result. Each link in the chain includes: the original capture (with EXIF, GPS, timestamp), the processing pipeline version, the model version used, the analyst (human or AI), and the output. Chains are append-only and tamper-evident — any modification breaks the signature. The Evidence Engine ensures that Landvex outputs are admissible in legal and regulatory contexts.

API endpoint that generates human-readable explanations for AI decisions, including visual attention maps and reasoning traces.

The /explain endpoint addresses the "black box" problem of AI. For any detection, classification, or scoring result, it produces: a saliency map (which pixels influenced the decision), a textual rationale ("classified as 'crack' because of linear dark region with edge discontinuity"), and a confidence breakdown (which features contributed how much). Explanations are essential for regulatory compliance, customer trust, and model debugging. The endpoint supports multiple explanation methods: LIME, SHAP, and integrated gradients.

API endpoint that pulls structured data from unstructured visual inputs — OCR, barcode reading, serial number recognition, and form field extraction.

The /extract endpoint transforms visual information into structured data. It performs OCR on documents and signs, reads QR codes and barcodes, recognises serial numbers and VINs, and extracts form fields from structured documents. It handles rotated, skewed, and low-contrast inputs through preprocessing pipelines. Output is JSON with normalised text, bounding boxes per field, and confidence per extraction.

Structured observations captured on location by verified Zoomers — the raw material of the Landvex intelligence pipeline.

Field Data is not just photos. It is a complete, verified record of physical reality at a specific place and time: geolocation (GPS coordinates with accuracy estimate), timestamp (when captured), media (images or video), and metadata (device, conditions, mission context). Field Data is quality-checked by AI before entering the dataset. It is the foundational input from which all AAMOS analysis and Landvex intelligence is derived. Without Field Data, there is no intelligence.

See AMOS Fraud.

An image embedded with GPS coordinates at capture time, proving where the photo was physically taken.

Geotagging writes latitude, longitude, and altitude into the image's EXIF metadata at the moment of capture. All quiXzoom submissions must be geotagged — this is enforced by the app. Submissions without GPS data are rejected before upload. Geotagging is the foundation of location verification: without it, there is no way to confirm the observation was captured at the correct mission location.

A structured capture protocol where AVO guides the Zoomer to collect specific evidence types from defined vantage points.

Guided Evidence is the quiXzoom capture mode for complex missions. Instead of free-form photography, the Zoomer follows a step-by-step protocol: "Photograph the north facade from 5 metres," "Capture the serial number plate," "Document the damage area with a reference object." AVO validates each step in real time and prevents submission until all required evidence is collected. Guided Evidence produces higher-quality, more consistent data than unstructured capture.

A cryptographic signature method used to verify that webhook payloads and API responses were genuinely sent by AAMOS and not tampered with in transit.

HMAC (Hash-based Message Authentication Code) with SHA-256 combines the payload with a shared secret key to produce a unique signature. AAMOS attaches this signature to every webhook request and sensitive API response. The receiving system recomputes the signature and compares — if they match, the payload is authentic and unmodified. This is industry-standard practice used by Stripe, GitHub, and other major API providers.

See AMOS Identity.

See AMOS Infrastructure.

A composite score (0–100) representing the overall risk level of an infrastructure asset based on visual condition, historical trends, and environmental factors.

The Infrastructure Risk Index aggregates multiple signals from AMOS analysis: surface condition scores, structural anomaly counts, change severity, age estimates, and environmental stress factors (freeze-thaw cycles, traffic load, proximity to water). A score of 0–30 indicates low risk (routine maintenance), 31–60 moderate risk (monitor closely), 61–80 high risk (plan intervention), and 81–100 critical risk (immediate action required). The index is recalculated with every new observation and includes confidence intervals.

See AMOS Inspection.

JavaScript Object Notation — the standard data format for all AAMOS API requests and responses.

JSON is a lightweight, human-readable data interchange format used throughout the AAMOS API. All endpoints accept JSON payloads and return JSON responses. Structured outputs include nested objects for complex results (e.g., a detection with bounding box, label, confidence, and child objects). GeoJSON (RFC 7946) is used for geographic data structures.

Control Intelligence company operating through Landvex AB (EU) and Landvex Inc. (US) — the decision layer of the ecosystem.

Landvex combines field-verified intelligence from quiXzoom, AI-driven analysis from AAMOS, and cross-market operational coverage through its European and North American entities. The Landvex platform enables infrastructure owners, governments, utilities, and enterprises to identify risks, opportunities, deficiencies, and emerging conditions before they become operational problems. Landvex does not sell AI — it sells intelligence based on AMOS engines. Key products: VIMS, Reality Alerts, RIOS, and Control Intelligence APIs.

The verification that a captured biometric (face, fingerprint) comes from a live person, not a photo, mask, or replay.

Liveness detection prevents spoofing attacks in identity verification. AAMOS Identity uses multiple liveness signals: challenge-response (blink, turn head, smile), texture analysis (detecting screen glare, paper edges, or mask seams), depth estimation (3D face geometry), and temporal consistency (natural micro-movements). A liveness score above the threshold is required for authentication to proceed. Liveness is critical for KYC, access control, and fraud prevention.

A defined data collection task on the quiXzoom platform — the atomic unit of field work.

A mission specifies what to observe, where to observe it, how to capture it, and by when. It includes: a geofence (GPS boundary), a time window, a capture protocol (number of photos, required angles), a compensation amount (in USD/QZ Token), and quality criteria. Missions are created by orderers (customers) and claimed by Zoomers (field collectors). When completed and approved, the mission triggers payout and delivers the observation package to the orderer.

The authentication protocol used by the AAMOS API — industry-standard delegated access without sharing passwords.

AAMOS uses OAuth 2.0 with bearer tokens for API authentication. Developers register applications to receive client credentials, then exchange them for short-lived access tokens. Refresh tokens enable long-lived integrations without storing passwords. Scopes control which engines and endpoints an application can access. OAuth 2.0 is the same protocol used by Google, GitHub, and Stripe.

A verified field record captured through quiXzoom — the fundamental unit of reality data.

An observation is a structured, verified record of physical reality at a specific place and time. Every observation includes: geolocation (GPS coordinates with accuracy estimate), timestamp (when captured), media (images or video), and metadata (device, conditions, mission context). Observations are quality-checked by AI before entering the dataset. They are the raw material from which all Landvex intelligence is derived.

Open Neural Network Exchange — the standard format for AAMOS model deployment, enabling cross-platform inference.

ONNX is an open standard for representing machine learning models. AAMOS trains models in PyTorch, then exports to ONNX for deployment. ONNX models run on multiple backends: NVIDIA GPUs (via TensorRT), Intel CPUs (via OpenVINO), mobile devices (via ONNX Runtime), and cloud instances. This flexibility allows AAMOS to deploy the same model to edge devices (quiXzoom app), on-premise servers (enterprise customers), and cloud infrastructure (high-throughput API) without retraining.

Physical Business Opportunity Intelligence — proactively identified commercial needs in the built environment, delivered as qualified leads.

PBOI is Landvex's highest-value product. Instead of waiting for customers to request observations, Landvex continuously monitors the built environment through the quiXzoom network and identifies business opportunities: dirty storefront windows (for cleaning services), faded facades (for painters), damaged parking signs (for maintenance contractors), vacant lots (for developers). Each opportunity is geotagged, photographed, severity-scored, and packaged as a qualified lead. PBOI transforms Landvex from a data vendor into a lead-generation platform.

AMOS engine that estimates the likely future evolution of asset conditions, risk levels, and maintenance needs.

The Prediction Engine combines historical observation trends, environmental data, material science models, and usage patterns to forecast future asset states. It answers questions like: "Given current crack propagation rates and freeze-thaw exposure, when will this bridge deck require repair?" Outputs include time-to-failure estimates, probability distributions for intervention windows, and scenario modelling (what-if traffic increases 20%?). Predictions include explicit uncertainty bounds — they are estimates, not guarantees.

The internal currency unit of the quiXzoom platform, used for mission pricing, Zoomer compensation, and platform accounting.

QX Credits are quiXzoom's internal accounting unit. All mission prices, Zoomer earnings, and platform transactions are denominated in QX Credits. 1 QX Credit = 1 USD (fixed exchange rate, no speculation). QX Credits provide a stable, platform-native unit of account that simplifies global operations — Zoomers in any country earn QX Credits, which are converted to their local currency at payout via Stripe Connect. Orderers purchase QX Credit packages to fund missions.

Blockchain-tracked token representing earned value on the quiXzoom platform — immutable, auditable, and non-speculative.

The QZ Token is a blockchain-based representation of quiXzoom earnings. Each token is created when a mission is approved and stored on an immutable ledger. The ledger provides complete auditability: every token's origin (which mission, which Zoomer, which timestamp), transfer history, and redemption record is permanently recorded. QZ Tokens are NOT a cryptocurrency for trading — they are an accounting and transparency mechanism. 1 QZ Token = 1 USD = 1 QX Credit.

Field Intelligence Network — the global crowdsourced data collection platform that feeds the Landvex ecosystem.

quiXzoom is a marketplace where orderers (customers) post data collection missions and Zoomers (verified field contributors) complete them for compensation. It is a standalone product and brand, not a module of AAMOS. quiXzoom uses AAMOS engines under the hood for AI review and quality control, but presents itself as an independent field intelligence platform. quiXzoom's purpose is to collect verified, geotagged, timestamped observations of physical reality at global scale.

Unprocessed visual captures from field operations — the input state before AI review, geotag validation, or structured packaging.

Raw Data refers to the initial media files captured by Zoomers before any processing: the original JPEG/HEIC from the camera, with native EXIF, before AI review, before quality scoring, before packaging into an observation. Raw Data is retained for a limited period for dispute resolution and audit purposes, then purged according to the data retention policy. Customers never receive Raw Data — they receive processed, verified observations.

Real-time notifications triggered when AAMOS detects significant changes, contradictions, or anomalies in monitored assets.

Reality Alerts are Landvex's proactive notification system. When the Change Engine or Risk Engine detects a condition that exceeds a customer's configured threshold, a Reality Alert is generated and delivered via webhook, email, SMS, or in-app notification. Alert types include: Contradiction Alert (unexpected change detected), Risk Escalation (index crossed a threshold), Coverage Gap (area hasn't been observed within SLA), and Opportunity Signal (PBOI lead identified). Each alert includes the evidence, confidence score, recommended action, and a link to the full analysis.

AMOS engine that verifies an image shows actual reality — not a manipulated, synthetic, or replayed scene.

The Reality Engine is AAMOS's first line of defence against fraudulent or manipulated inputs. It analyses images for signs of tampering: inconsistent lighting, unnatural shadows, compression artefacts at edit boundaries, cloned regions, and AI-generated content (deepfakes). It also verifies EXIF integrity, GPS consistency, and timestamp plausibility. The Reality Engine does not determine what the image shows — it determines whether the image is trustworthy. A failed Reality Engine check blocks the observation from entering downstream analysis.

The measurable difference between reported or planned reality and observed reality — Landvex's core detection target.

The Reality Gap is the space between what should be and what is. It manifests as: infrastructure that is reported maintained but observed deteriorated, buildings that are permitted as safe but show structural concerns, or urban plans that describe green spaces but deliver concrete. Landvex's Contradiction Analysis systematically identifies Reality Gaps by comparing official records, design specifications, and historical baselines against current observations. The Reality Gap Index (RGI) quantifies this gap per asset or region.

A measurable, verifiable observation that describes a specific aspect of the physical world.

A Reality Signal is any observation that carries information about the state of physical reality: a photo of a cracked bridge deck, a GPS-tagged image of flood damage, a timestamped record of a building's facade. Reality Signals are distinguished from synthetic or inferred data by their direct connection to physical measurement — a Zoomer was there, the camera captured light from the actual scene, the GPS recorded the actual coordinates. Reality Signals are the antidote to assumptions and desk-based estimates.

Service Level Agreement governing how frequently observations must be updated to remain current — a core component of PBOI subscriptions.

The Refresh SLA defines the maximum age of observations for a given asset class and geographic area. Different asset types have different refresh requirements: dirty storefront windows need weekly refresh (conditions change fast), while building facades can be quarterly (deterioration is slower). The Coverage Map visualises SLA compliance: green = within SLA, yellow = approaching expiry, red = stale. Refresh SLA is a contractual commitment in Landvex subscription agreements.

Reality Intelligence Operating System — Landvex's platform for orchestrating data collection, analysis, and decision delivery.

RIOS is the operating system that coordinates the entire Landvex intelligence pipeline. It manages: mission scheduling and dispatch to quiXzoom, data ingestion from field operations, AMOS engine orchestration, index computation, alert generation, and customer dashboard delivery. RIOS is not a product sold separately — it is the infrastructure that powers all Landvex intelligence products. Think of it as the "OS" that runs on top of the quiXzoom + AAMOS stack.

AMOS engine that transforms observations into risk scores, probability estimates, and prioritised action lists.

The Risk Engine ingests observation data (condition scores, anomaly counts, change severity) and combines it with contextual factors (asset age, material type, environmental exposure, usage intensity) to produce risk assessments. It outputs: per-asset risk scores, portfolio-level risk distributions, time-to-intervention estimates, and cost-weighted priority rankings. The Risk Engine supports multiple risk models: structural failure, insurance claim likelihood, maintenance cost escalation, and safety incident probability.

See AMOS Safety.

API endpoint that assigns a numerical score to an image, object, or asset based on trained quality, risk, or condition models.

The /score endpoint produces calibrated numerical assessments. Unlike /classify (which assigns categories), /score returns a continuous value on a defined scale. Score types include: condition score (0–100, where 100 = perfect), risk score (0–100, where 100 = critical), quality score (0–100, where 100 = publication-ready), and custom scores trained for specific domains. Each score includes confidence intervals and the contributing feature breakdown.

API endpoint that performs instance segmentation — identifying the exact pixel boundaries of objects within an image.

The /segment endpoint goes beyond /detect (bounding boxes) to produce pixel-accurate masks for each detected object. Segmentation enables precise area measurement (e.g., "crack covers 0.34 m²"), damage quantification, and detailed visual analysis. It supports both semantic segmentation (class per pixel) and instance segmentation (individual object masks). Output includes polygon coordinates, mask area, and per-pixel confidence.

AI-guided capture mode that automatically detects when sufficient visual evidence has been collected and signals completion.

Smart Scan is the entry-level AVO mode. As the Zoomer captures images, AMOS Vision analyses them in real time and provides feedback: "Damage detected — capture closer," "Insufficient lighting — retry," "Object complete — move to next." When the sufficiency threshold is reached (trust_achieved ≥ trust_threshold), Smart Scan signals completion with a green indicator. This removes guesswork from field capture and ensures consistent evidence quality across different Zoomers and conditions.

The determination that enough evidence has been collected to support a verified claim — calculated per-claim, never globally.

Sufficiency is AVO's core decision function. For each verified_claim, the system accumulates evidence quality signals: image sharpness, coverage of required angles, presence of key features, liveness confirmation, and cross-validation consistency. The sufficiency formula is: trust_achieved = Σ (weight_i × confidence_i × liveness_factor). When trust_achieved ≥ trust_threshold, the claim is marked sufficient. The threshold is set per knowledge pack and per claim type — never globally. This ensures that a passport verification requires different evidence than a facade inspection.

High-urgency mission deployment after acute events — storms, floods, accidents — offering elevated compensation for rapid documentation.

Surge events are time-critical mission activations triggered by disasters or emergencies. They carry 2–5× standard compensation to incentivise rapid response. Surge alerts are pushed as priority notifications to all Zoomers in the affected zone, with Professional tier Zoomers notified first. Surge missions typically have tighter specifications and stricter AI review given their legal and claims significance. They are the primary mechanism for post-disaster damage documentation.

An observation or image that carries a precise capture time, proving when the data was collected.

Timestamps are embedded in image EXIF metadata at capture time and cross-validated by server-side ingestion timestamps. They prove that an observation was made within a required time window — critical for insurance claims, regulatory compliance, and legal evidence. Timestamp verification includes: EXIF timestamp vs. server time drift check (rejecting images with suspicious future or past dates), mission time window compliance, and chain-of-custody timestamping on the immutable ledger.

API endpoint that follows an object or asset across multiple observations over time, maintaining identity and building a temporal history.

The /track endpoint enables longitudinal analysis by associating observations of the same asset across time. It uses visual features, GPS proximity, and metadata matching to establish identity persistence. Output includes: a timeline of observations, condition trend graphs, change event logs, and predictive trajectories. /track is essential for infrastructure monitoring, where understanding deterioration rates requires comparing the same asset across quarters or years.

A composite metric (0–100) evaluating the overall condition, safety, and functionality of an urban area based on multi-source observations.

The Urban Intelligence Score aggregates multiple sub-indices into a single comparable metric: Infrastructure Risk Index, Sanitation Index, Safety Score, Commercial Vitality, and Environmental Quality. It enables city-to-city comparison, neighbourhood ranking, and trend tracking over time. The score is computed from quiXzoom observations, AMOS analysis, and external data sources (weather, traffic, public records). It is a key output for municipal customers and urban planners.

A cryptographically signed, immutable assertion about physical reality — the unit of truth in the AAMOS evidence system.

A Verified Claim is not the image, not the evidence package, but the assertion itself: "Bridge 2847 has a crack 43cm long on the north facade, confidence 94%, observed 2026-07-12 09:47 UTC by Zoomer Z-2847." Each claim is signed with Ed25519, stored append-only, and includes provenance (who observed, what model analysed, what human reviewed). Corrections create new claims with backward references — the original remains immutable. Verified Claims are the currency of trust in the Landvex ecosystem.

API endpoint that checks whether an image, document, or asset meets specified criteria — the general-purpose validation interface.

The /verify endpoint is a flexible validation tool. It accepts criteria definitions ("must contain a face", "must show a readable serial number", "must be captured after 2026-01-01") and returns a pass/fail result with detailed reasoning. It is used for document verification, quality gate checks, compliance validation, and custom business rules. Criteria can be predefined (from knowledge packs) or custom (supplied in the API request).

Visual Infrastructure Monitoring System — Landvex product for continuous, AI-powered monitoring of critical infrastructure assets.

VIMS applies the AAMOS engine stack to critical infrastructure: bank ATMs, EV charging stations, traffic signals, utility meters, and safety equipment. It creates a visual baseline per object, then compares new observations against that baseline to detect tampering, damage, wear, or unauthorised modifications. Each anomaly is risk-classified (Green/Yellow/Orange/Red) and triggers appropriate alerts. VIMS combines quiXzoom field collection with AMOS analysis and Landvex decision delivery into a complete monitoring solution.

See AMOS Vision.

An automatic HTTP POST notification sent by AAMOS when a defined event occurs — enabling real-time integration without polling.

Webhooks enable event-driven integration between AAMOS and customer systems. When a batch analysis completes, a model is deployed, or an alert threshold is crossed, AAMOS sends a JSON payload to the customer's configured webhook URL. The payload includes event type, timestamps, result summaries, and links to full results. Customers verify webhook authenticity using HMAC-SHA256 signature verification. Undelivered webhooks are retried with exponential backoff over 24 hours.

You Only Look Once — the real-time object detection architecture used as the foundation for AAMOS Vision models.

YOLO is a family of deep learning models designed for fast, accurate object detection in images and video. AAMOS uses YOLOv8 and YOLOv9 variants, trained on domain-specific datasets (infrastructure, safety, fraud), and exported to ONNX for cross-platform deployment. YOLO's single-pass architecture enables real-time inference: a typical AAMOS model processes a 1920×1080 image in 15–40ms on a modern GPU. This speed is essential for real-time AVO guidance and high-throughput batch processing.

A verified field data collector on the quiXzoom platform — an individual who claims missions, captures on-location photography, and earns compensation per approved submission.

Zoomers are the supply side of the quiXzoom marketplace. Anyone with a compatible smartphone, a willingness to be verified (KYC via AMOS Identity), and a linked bank account can become a Zoomer. Zoomers work independently, on their own schedule, within their chosen zone. They earn per approved submission — there is no hourly pay, no minimum hours, and no obligation. Zoomers progress through tiers (Supplementary → Active → Professional) as they build volume and quality.