Causal AI for time series

Understand what drives your data, and what to change.

Kausyn brings causal AI to the time series behind your decisions — separating the real drivers from correlations that don't hold, across finance, mobility, healthcare, energy and beyond.

Causal structure · recovered from your signals
X₁ X₂ X₃ X₄
a directed driver, distinguished from mere correlation
Why causal

Correlation tells you what moved together. Causality tells you what to change.

The problem

Standard analytics reads patterns. It doesn't read cause.

Two gaps sit between raw time series and a decision you can defend. Kausyn is built to close both.

Gap 01

Correlation is not a driver

Variables that move together look identical whether one causes the other or a hidden factor moves both. Models that exploit the pattern give confident answers that fail when conditions change — and offer no lever to act on.

Gap 02

Conditions rarely stay fixed

Correlations spike in a crisis, demand shifts across seasons, physiology changes with treatment. Tools that assume a fixed structure blur these regimes into one misleading average. The change itself carries information — most methods discard it.

The technology

Causal discovery for the time series behind your decisions.

Kausyn recovers the direction and structure of cause and effect from observational time series — and holds up in the real-world conditions where data drifts and rules shift.

Finds cause, not correlation

Recovers directed cause-and-effect structure from observational data alone — no controlled experiments or interventional trials required.

Holds up when things move

Built to stay reliable through drift, regime shifts, and changing volatility — the real-world instability that quietly defeats conventional methods.

Reports only what holds

Every recovered relationship passes statistical error control, so a link is stated when the evidence supports it — not because a number looked large.

How it works

From raw signals to a graph you can act on.

STEP 01

Bring your series

Multivariate observational time series go in as recorded — no need to pre-clean them into an artificially stable shape first.

STEP 02

Discover the structure

The method reads how each variable behaves over time to resolve which variables drive which, and in what direction.

STEP 03

Return a validated graph

Out comes a directed causal graph with error control applied — the drivers behind the system, and where to focus.

Sectors

One capability, many domains.

Causal discovery is horizontal. Anywhere decisions rest on time series, the same question applies: what is actually driving this. These are the sectors Kausyn is focused on.

Finance

Markets & risk

Lead–lag structure and exposure between instruments and institutions through shifting correlations.

Mobility

Transport & logistics

Drivers of demand, delay, and flow across networks, where load and behaviour change constantly.

Health

Medical & clinical

Physiological couplings and treatment effects read from monitored signals over time.

Energy

Energy & climate

Relationships that move with seasons and events, from grid load to environmental drivers.

Retail

Demand & marketing

What truly moves demand and response, past the correlations a dashboard would report.

Industry

Operations & sensors

Root drivers behind quality, throughput, and failure in sensor-rich industrial processes.

The company

Research-driven, and built method-first.

Kausyn is an early-stage causal-AI company based in Morocco. We develop our core methods ourselves and prove them against established benchmarks before putting them to work on real systems. The near-term focus is turning that foundation into results on real-world data, one sector at a time.

FocusCausal discovery from time series
FoundationPeer-reviewed research · under review
BenchmarksValidated
Real-world validationIn progress
Performance to date is measured on established benchmark datasets. Results on real-world data are being completed and will be published as they become available.
Contact

Have time-series data where cause matters?

We work with data partners, collaborators, and research groups. The best place to start is a real dataset where getting cause and effect right is worth it.

Morocco