NewResearch Architect is live

Data in. Decisions out.

Entanglix builds data-driven SaaS products and a growing collection of AI agents — turning research questions, climate risk and environmental data into answers you can defend.

Explore the productsRequest a demo
Research Architect — influence graph
Urban heatTree canopyImpervious surfaceNight-time temp.Heat illnessIncomeAC access
DrivingNecessaryModifying
Products

Three products. One way of thinking.

Each one takes messy real-world evidence, reasons over it with AI, and hands back an answer with its uncertainty attached.

AI research copilot · SaaS

Research Architect

Design research, not just run it.

The hardest part of research is rarely the analysis — it is deciding what to ask. Research Architect turns a question into an interactive influence graph you can argue with: a shared reasoning space where researchers and AI frame hypotheses together.

  • Map key factors, mechanisms and relationships
  • Integrate literature and expert knowledge
  • Evaluate research scope against real data coverage
  • Develop grounded, testable hypotheses — with evidence and uncertainty visible
Tri-projection
System → graph → scope → data
Human-in-the-loop
Researchers correct the AI, not the reverse
Open the appresearcharchitect.ai
Research Architect — influence graph
Urban heatTree canopyImpervious surfaceNight-time temp.Heat illnessIncomeAC access
DrivingNecessaryModifying
Climate risk intelligence · Banks & insurers

FloodVuln Global

Hazard models tell you where the water goes. We tell you what it does.

A vendor-neutral vulnerability layer that converts flood depth into a full probability distribution of physical damage — specific to building archetype, geography and the local evidence you can actually produce. Bring depths from any hazard vendor; the vulnerability layer is the product.

  • Underwriting, pricing, accumulation, model validation and claims triage on one versioned engine
  • Global-to-local calibration, from a global prior down to a carrier's own portfolio history
  • Aleatory and epistemic uncertainty reported apart — insurance-grade
  • API and portfolio batch delivery with reproducible model versions
1.86M
flood claims behind the fit
4,483
archetype × region units calibrated
270
published curve points harmonized
Explore FloodVulnRequest a pilot
FloodVuln Engine — single risk
0.000.250.500.751.00mean
0.55 m
Mean DR0.184
P10 – P90.03–.49
P(DR > 50%)9.5%
GradeG3
Environmental data service · Public sectorIn collaboration with UNDP

GH-PM25 Observatory

Daily 1-km fine particulate matter for Ghana.

A national air-quality surface where ground monitors are sparse. We fuse satellite, reanalysis, land-use and population data with the monitor network to estimate PM2.5 for every square kilometre, every day — with honest uncertainty and direct comparison to health standards.

  • Population-weighted exposure and exceedance against WHO and Ghana EPA limits
  • 90% prediction intervals on every estimate
  • Low-cost sensor calibration and monitor-network assimilation
  • Validated leave-city-cluster-out, so it holds where no monitor exists
1 km
daily national grid
10
open data sources fused
455 × 650
cells, coast to Sahel
Open the ObservatoryBring it to your country
GH-PM25 Observatory — national surface
TamaleKumasiAccra
PM2.5 · µg/m³
WHO 15EPA 35
Seasonal cycle
JAJOD
Harmattan peak, Dec – Feb
Daily1 km90% PI
Platform

One stack under everything. Data at the base, agents on top.

Data
Models
Agents
03

Agents

Specialised AI agents that frame questions, challenge assumptions and explain results — with a human in the loop wherever judgement matters.

02

Models

Hierarchical, calibrated, uncertainty-aware models. Every prediction carries its confidence and reproduces from a pinned version.

01

Data

Satellite, reanalysis, claims, sensors, literature and expert knowledge — harmonized into one evidence layer with full provenance.

Uncertainty is a feature.

We report what the model knows and what it does not — separately. No silent extrapolation.

Evidence you can audit.

Sources, versions, rejected alternatives and expert priors are recorded, not remembered.

Humans stay essential.

Agents propose; people correct assumptions, add domain knowledge and choose the direction.

Agent collection

Agents that know what they don't know.

AI agents should not propose blindly, grow overconfident on incomplete evidence, or abandon good ideas too early. Ours are specialised, grounded in data, and built to be corrected.

Frames the question

Hypothesis Architect

Turns a research question into an influence graph, then proposes hypotheses the graph and the data can actually support.

Runs inside Research Architect
Grounds every edge

Literature Scout

Finds what the literature supports, what it contests, and where it is silent.

Runs inside Research Architect
Checks scope against data

Coverage Auditor

Tells you what your study can detect, what your data can measure — and what got left out.

Runs inside Research Architect
Prices physical damage

Vulnerability Engine

Converts flood depth into a calibrated damage distribution for any archetype and region, and flags out-of-domain requests instead of guessing.

Runs inside FloodVuln Global
Learns from ten observations

Local Calibrator

Updates global priors with sparse local claims and structured expert evidence — by exactly as much as it should.

Runs inside FloodVuln Global
Fills the gaps between monitors

Exposure Mapper

Fuses satellite, reanalysis and ground sensors into a daily 1-km pollution surface with prediction intervals.

Runs inside GH-PM25 Observatory
Data services

Delivered the way your team already works.

Subscribe to an app, pull from an API, or have us stand up a data service for your region and your risk.

APIs

Versioned endpoints that return distributions, confidence and provenance — not a bare number.

Data products

Analysis-ready gridded surfaces and portfolio batches, refreshed on a schedule you can rely on.

Hosted apps

Browser-based SaaS workspaces and observatories. Nothing to install, nothing to maintain.

Custom builds

New geographies, new hazards, new agents — delivered on the same stack, with the same rigour.

POST /v1/vulnerability200 OK
{
  "archetype": "US.RES.SF.WOOD.1STORY.SLAB",
  "depth_m": 0.55,
  "damage_ratio": {
    "mean": 0.184,
    "p10": 0.03,  "p90": 0.49,
    "p_exceed_50": 0.094
  },
  "calibration_level": "G3",
  "in_domain": true,
  "model_version": "1.0.0"
}

Illustrative response from the FloodVuln engine.

1.86M
insurance claims modelled
1 km
daily national air-quality grid
4,483
locally calibrated risk units
10+
open data sources fused
Company

Entanglix is named for quantum entanglement — two things that cannot be described apart. For us, those are data and intelligence. We build software where neither works without the other.

01

Banks & lenders

Climate-risk views on collateral and portfolios.

02

Insurers & reinsurers

Vulnerability, pricing and model validation.

03

Development agencies

National environmental data where monitoring is sparse.

04

Research institutions

Human–AI collaboration for study design.

Contact

Let's put your data to work.

Request a demo, start a pilot, or tell us about a problem none of our products solves yet.

Email
info@entanglix.tech
Web
entanglix.tech
Based in
Atlanta, Georgia, USA