---
title: Kausalyze | Causal AI for process manufacturing. Know why your plant fails, not just when.
description: Find the upstream cause of recurring plant problems, validate the evidence with engineers and determine what to change. Start with a 12 to 16 week pilot on one unit.
---

[![Kausalyze](https://www.kausalyze.com/hubfs/raw_assets/public/kausalyze-rebuild/logo.svg)](https://www.kausalyze.com/?hsLang=en)

- [Home](https://www.kausalyze.com/?hsLang=en)
- [How it works](https://www.kausalyze.com/how-it-works?hsLang=en)
- Company 
  
    - [About](https://www.kausalyze.com/about?hsLang=en)
    - [Contact](https://www.kausalyze.com/contact?hsLang=en)
    - [Careers](https://www.kausalyze.com/careers?hsLang=en)
- Resources 
  
    - [Insights](https://www.kausalyze.com/kausalyze-ltd-blog/tag/insights?hsLang=en)
    - [News](https://www.kausalyze.com/kausalyze-ltd-blog/tag/news?hsLang=en)

[Book a technical review](https://www.kausalyze.com/book-demo?hsLang=en)

Causal AI for process manufacturing

# Know why your plant fails. *Not just when.*

Predictive maintenance flags symptoms or forecasts failure. Kausalyze builds a causal model of your process to trace the problem to its upstream driver, show the evidence and identify what to change.

[Book a technical review](https://www.kausalyze.com/book-demo?hsLang=en) [See how it works](https://www.kausalyze.com/how-it-works?hsLang=en)

+0 +0 +1 +1 +0 +3 +0 +3 +0 A\_LEAN\_FLOWLean amine flow A\_INLET\_TEMPFlue gas inlet temp A\_ABS\_TEMPAbsorber bed temp A\_RICH\_LOADRich amine loading U\_STEAM\_PRESLP steam header A\_CO2\_CAPCO2 capture rate R\_REB\_DUTYReboiler steam duty L\_DEGRAD\_PRODDegradation products +2.4σin band Root cause set −2.1% +0 +1 +0 +3 A\_LEAN\_FLOWLean amine flow A\_ABS\_TEMPAbsorber bed temp A\_RICH\_LOADRich amine loading A\_CO2\_CAPCO2 capture rate R\_REB\_DUTYReboiler steam duty Root cause set −2.1% +2.4σin band

AttributionR\_REB\_DUTY · share of the fault explained

A\_LEAN\_FLOW**0.41**

A\_ABS\_TEMP**0.23**

U\_STEAM\_PRES**0.15**

A\_RICH\_LOAD**0.12**

- University of Sheffield spinout
- Backed by Innovate UK
- Plug and Play Tech Center
- Greentown Labs
- Founded by chemical engineers

The problem

## Billions spent on AI.

## Downtime costs keep rising.

Process manufacturers have invested heavily in data infrastructure and predictive tools. Those tools have become better at flagging problems, but an alert still does not tell you what caused the problem or what to change. Engineers must make that connection while experienced colleagues retire and recurring failures keep costing the plant money.

11%

**of revenue lost to unplanned downtime**across the Fortune Global 500. Up from 8% in 2019.

+62%

**rise in the cost of downtime since 2019**even as incident frequency has fallen.

~$129M

**estimated annual downtime loss**for a single large process facility

Sources: Siemens, The True Cost of Downtime 2024. Rockwell, State of Smart Manufacturing 2026. Per-facility figure is a benchmark-derived estimate for modelling and varies by site.

Where the money goes

### Every day without the cause costs money

A process upset does not pause while the team investigates. Finding the true root cause of a recurring problem takes days or weeks. Lost yield, off-spec product, excess energy and avoidable maintenance add up the whole time.

### An alert does not tell you what to change

A warning helps you plan a response. Resolving a recurring problem also requires the upstream cause, evidence your engineers can inspect and a specific change with an expected impact.

What engineers can do

## One causal model. Better operating decisions.

Assess process health, find root causes, forecast time to a limit, test interventions and prescribe changes, all using the same causal model.

01

### Process Health

Assess equipment condition and process stability across the connected unit. See a change in the context of the upstream variables driving it.

Indicated value

35 to 45%

less downtime for plants on condition-based maintenance programmes. Maintenance cost down 25 to 30%.

US DOE, O&M Best Practices Guide

02

### Causal Root Cause

Trace a recurring failure or quality problem to its upstream drivers. Inspect the causal path and supporting evidence with your engineers.

Indicated value

$1B+ / yr

lost to preheat-train fouling in US refining alone. One unit can lose millions a year.

Heat Transfer Engineering 2024; Müller-Steinhagen

03

### Failure Forecasting

Estimate when a condition is likely to cross a limit, with confidence levels and supporting evidence, so your team can plan an intervention.

Indicated value

$500k to $2M

per hour of unplanned downtime in heavy process industry. An unplanned outage costs around 50% more than the same work planned.

Siemens, True Cost of Downtime 2024; industry estimate

04

### Test and prescribe changes

Test a set-point change in the model before touching the plant. Compare interventions and quantify the expected effect on production, energy or maintenance.

Indicated value

+10% output

and 25% less high-pressure steam at a polyurethane plant, from set-point changes alone. No capex.

McKinsey, Digital in chemicals

Indicated values are published industry figures, not Kausalyze performance claims. Site-level value is established in the pilot, against the plant's own records.

The Product

## From process health to a prescribed change

01Monitor

02Diagnose

03Forecast

04What if

05Optimise

Causal graph · amine CO2 capture unit 8 variables · lag in hours

+0 +0 +1 +1 +0 +3 +0 +3 +0 A\_LEAN\_FLOWLean amine flow A\_INLET\_TEMPFlue gas inlet temp A\_ABS\_TEMPAbsorber bed temp A\_RICH\_LOADRich amine loading U\_STEAM\_PRESLP steam header A\_CO2\_CAPCO2 capture rate R\_REB\_DUTYReboiler steam duty L\_DEGRAD\_PRODDegradation products In band +2.4σ Time to limit back in band Root cause try −2.1% set −2.1% in bounds in bounds near limit +0 +1 +0 +3 A\_LEAN\_FLOWLean amine flow A\_ABS\_TEMPAbsorber bed temp A\_RICH\_LOADRich amine loading A\_CO2\_CAPCO2 capture rate R\_REB\_DUTYReboiler steam duty In band +2.4σ Time to limit back in band Root cause try −2.1% set −2.1% in bounds in bounds near limit

AttributionR\_REB\_DUTY · share of the fault explained

A\_LEAN\_FLOW**0.41**

A\_ABS\_TEMP**0.23**

U\_STEAM\_PRES**0.15**

A\_RICH\_LOAD**0.12**

R\_REB\_DUTY · Reboiler steam duty, MW In band

Operating band Limit 14.0 MW 14.0 13.0 12.0 11.0 MW −2 h −1 h now Do nothing A\_LEAN\_FLOW −2.1% Time to limit Departure set A\_LEAN\_FLOW −2.1%

Step 1 of 5

### Monitor

Assess each variable alongside its upstream drivers to identify changes in process health across connected equipment.

Step 2 of 5

### Diagnose

Trace a fault back through the causal relationships to the upstream driver, with evidence your engineers can review.

Root cause: A\_LEAN\_FLOW

Step 3 of 5

### Forecast

Estimate the time to a process limit and review the forecast evidence to plan when to intervene.

Step 4 of 5

### What if

Test the change before you touch the plant. Compare a proposed set-point change with leaving conditions unchanged, and inspect the predicted downstream effects.

try A\_LEAN\_FLOW −2.1%

Step 5 of 5

### Optimise

Review the prescribed change and its expected impact on the unit. Quantify the value before deciding whether to implement it through your existing operating procedures.

set A\_LEAN\_FLOW −2.1%

Why it's different

## Autonomous manufacturing needs a causal backbone.

 Predictive tools flag anomalies. AI co-pilots search historical events. Kausalyze identifies the root cause, shows the evidence, and helps optimise the process, transforming data into decisions and saving real money.

 Constrained by physics, validated by engineers, and supported by evidence, Kausalyze provides the causal foundation for autonomous manufacturing.

| Capability | Predictive maintenance | AI co-pilots | Kausalyze |
| --- | --- | --- | --- |
| Finds the upstream cause Not just the symptom | ◑ Usually asset-level | ✕ | ✓ |
| Models cause across the process Equipment, units and shared utilities | ✕ Typically asset-by-asset | ✕ | ✓ |
| Shows the evidence behind the answer | ◑ | ◑ Depends on source retrieval | ✓ Causal path and evidence per link |
| Constrained by process physics | ◑ Varies by tool | ✕ No process model by itself | ✓ Physics and process connectivity |
| Engineers validate causal links | ◑ Review alerts and forecasts | ◑ Review generated answers | ✓ Confirm or reject important links |
| Tests what happens if you make a change | ✕ | ✕ | ✓ |
| Prescribes what to change | ◑ Usually maintenance action | ◑ Can suggest options | ✓ With expected downstream impact |
| Starts with existing historical data First validated root causes in 12 to 16 weeks | ◑ Requirements vary | ◑ Requirements vary | ✓ |

✓ Core capability ◑ Partial or implementation-dependent ✕ Not provided by the capability alone

 Capabilities vary by product and implementation. The comparison describes typical uses.

Industrial proof

## Validated on real plant data, with the engineers who run it.

Kausalyze is working with Huntsman Polyurethanes' Global Excellence Team on heat-exchanger fouling, building causal models of how operating conditions actually drive fouling and degradation, and validating them jointly with Huntsman's engineers.

![Huntsman](https://www.kausalyze.com/hs-fs/hubfs/raw_assets/public/kausalyze-rebuild/huntsman-logo-transparent.png?width=3755&height=1557&name=huntsman-logo-transparent.png)

[Read about the collaboration](https://www.kausalyze.com/kausalyze-ltd-blog/how-kausalyze-is-helping-huntsman-reduce-heat-exchanger-fouling?hsLang=en)

“

> Kausalyze gives us deeper insight into asset behaviour than conventional predictive maintenance tools.

Process Improvement Global Excellence Team, Huntsman Polyurethanes on the collaboration.

The pilot

## One unit. One recurring problem. 12 to 16 weeks.

Scope the problem, validate the cause with engineers and quantify the value of a prescribed change. For the pilot: no new sensors, no new control software and no live-system integration required.

Phase 1

### Scope the unit and problem

Choose one unit and a recurring, costly failure, quality problem or process instability. Agree the scope and data handoff.

**You get**Agreed scope and a data quality check

Phase 2

### Build the causal model

Build a model of the unit from historical plant data and process information.

**You get**A causal map of the unit with evidence per link

Phase 3

### Validate root causes with engineers

Review the important relationships and root-cause findings with your process engineers.

**You get**Ranked, validated root causes for the target problem

Phase 4

### Prescribe changes and quantify value

Identify interventions, estimate their impact and build the case for whether and where to expand next.

**You get**Prescribed changes and quantified expected value

[See the pilot phases and deliverables](https://www.kausalyze.com/how-it-works#pilot)

Calculator

## Downtime is a dollar problem

Enter your annual downtime exposure per site, the share you believe is avoidable once the causes are known, and the number of sites in scope.

Annual downtime exposure per site $129M

$20M$300M

Share you believe is avoidable 30%

10%50%

Facilities in scope 1

125

Illustrative gross annual value

$38.7M

Exposure × avoidable share × facilities

Illustrative only, not a performance guarantee. Before software and implementation costs. Default exposure is a benchmark-derived estimate for a single large process facility (Siemens, True Cost of Downtime 2024).

## See what's really driving your downtime.

Bring one recurring plant problem. Review the unit, available data and potential value with our engineers, and define the scope for a pilot.

[Book a technical review](https://www.kausalyze.com/book-demo?hsLang=en)

[![Kausalyze](https://www.kausalyze.com/hubfs/raw_assets/public/kausalyze-rebuild/logo.svg)](https://www.kausalyze.com/?hsLang=en)

Causal AI for process manufacturing. Founded by chemical engineers.

#### Product

- [Home](https://www.kausalyze.com/?hsLang=en)
- [How it works](https://www.kausalyze.com/how-it-works?hsLang=en)
- [Book a technical review](https://www.kausalyze.com/book-demo?hsLang=en)
- Cost Calculator

#### Company

- [About](https://www.kausalyze.com/about?hsLang=en)
- [Contact](https://www.kausalyze.com/contact?hsLang=en)
- [Careers](https://www.kausalyze.com/careers?hsLang=en)

#### Resources

- [Insights](https://www.kausalyze.com/kausalyze-ltd-blog/tag/insights?hsLang=en)
- [News](https://www.kausalyze.com/kausalyze-ltd-blog/tag/news?hsLang=en)

© 2026 Kausalyze Ltd. [Privacy](https://www.kausalyze.com/privacy?hsLang=en)

```json
{
  "@context" : "https://schema.org",
  "@id" : "https://www.kausalyze.com/#organization",
  "@type" : "Organization",
  "address" : {
    "@type" : "PostalAddress",
    "addressCountry" : "GB",
    "addressLocality" : "Nottingham",
    "addressRegion" : "Nottinghamshire"
  },
  "alternateName" : "Kausalyze",
  "award" : "IChemE Global Awards Finalist 2024",
  "contactPoint" : [ {
    "@type" : "ContactPoint",
    "areaServed" : [ "GB", "US", "SG", "AE" ],
    "availableLanguage" : "English",
    "contactType" : "sales",
    "url" : "https://www.kausalyze.com/contact"
  } ],
  "description" : "Kausalyze is a causal AI platform for process manufacturers. It finds the root causes of unplanned downtime that predictive maintenance misses in continuous process operations.",
  "founder" : {
    "@type" : "Person",
    "alumniOf" : {
      "@type" : "EducationalOrganization",
      "name" : "University of Sheffield"
    },
    "jobTitle" : "Chief Executive Officer",
    "name" : "Louis Allen"
  },
  "foundingDate" : "2023",
  "foundingLocation" : {
    "@type" : "Place",
    "name" : "Sheffield, United Kingdom"
  },
  "industry" : "Industrial Software / Manufacturing Analytics",
  "knowsAbout" : [ "Causal AI", "Causal inference", "Root cause analysis", "Process manufacturing", "Unplanned downtime reduction", "Industrial data analytics", "Predictive maintenance" ],
  "logo" : {
    "@type" : "ImageObject",
    "height" : 512,
    "url" : "https://www.kausalyze.com/hubfs/kausalyze-logo-1.png",
    "width" : 512
  },
  "memberOf" : [ {
    "@type" : "Organization",
    "name" : "NEPIC (North East of England Process Industry Cluster)",
    "url" : "https://www.nepic.co.uk/"
  }, {
    "@type" : "Organization",
    "name" : "Greentown Labs",
    "url" : "https://greentownlabs.com/"
  } ],
  "name" : "Kausalyze Ltd",
  "parentOrganization" : {
    "@type" : "EducationalOrganization",
    "name" : "University of Sheffield",
    "url" : "https://www.sheffield.ac.uk/"
  },
  "sameAs" : [ "https://www.linkedin.com/company/kausalyze" ],
  "url" : "https://www.kausalyze.com/"
}
```

```json
{
  "@context" : "https://schema.org",
  "@id" : "https://www.kausalyze.com/#website",
  "@type" : "WebSite",
  "inLanguage" : "en-GB",
  "name" : "Kausalyze",
  "publisher" : {
    "@id" : "https://www.kausalyze.com/#organization"
  },
  "url" : "https://www.kausalyze.com/"
}
```