Goodwood House builds inference engines that make invisible risk visible. Whether the subject is a supply chain, a sensor network, a climate subsystem or an eradication campaign, the method is the same: assemble the evidence into a formally structured knowledge graph, verify every claim against its provenance, and compute the quantities that decision-makers actually need — confidence, exposure and early warning rather than raw data.
Discuss an engagementMost analytical failures are not failures of data collection but failures of integration. The signal is there, distributed across systems, formats and organisational boundaries. What is missing is the machinery to bring it together with its provenance intact and to reason over it honestly. That machinery is what Goodwood House builds.
The core of the practice is a family of knowledge-graph and network-inference tools developed and proven across commercial, government and scientific problems. The foundation is the Ontology Auto-Tuner, a refinement engine for OWL and SHACL knowledge graphs demonstrated at a scale of 1.4 million triples. Around it sit applications in supply-chain exposure mapping, provenance-assured data verification, statistical early warning for complex systems, and quantified proof-of-absence for field campaigns.
Clients see the same practices whatever the domain. This is the working style of a practice built for environments — from government security programmes to peer-reviewed science — where being wrong quietly is more expensive than being wrong openly.
Analyses are pre-registered before data is touched, so that a positive result means what it appears to mean.
Negative findings are reported as plainly as positive ones, because a method whose failures are hidden cannot be trusted where it succeeds.
Every claim, including machine-generated content, is verified against a source before it is used.
Directions that the evidence does not support are closed cleanly rather than kept alive for appearances.
The practice's core work: knowledge-graph engines, network inference and provenance-assured verification applied to problems where the risk is real, distributed and not yet quantified.
A knowledge graph is only as useful as its consistency, and at real-world scale, consistency does not survive contact with real data unless something enforces it. The Ontology Auto-Tuner is an automated refinement system for OWL ontologies and SHACL constraint shapes that detects structural weaknesses, proposes corrections and validates the result.
The Auto-Tuner is the common ancestor of everything else on this page. It is what allows a supply chain, a sensor estate or an evidence base to be represented formally enough that inference over it can be trusted, and it is available both as background IP within consulting engagements and as the basis of licensed applications.
DependMap maps multi-tier supply-chain dependency risk. A bill of materials is ingested, the supplier network is traversed beyond the first tier, and exposures that are invisible in any single procurement record become explicit — the canonical example being indirect rare-earth exposure that surfaces only at the second and third tier of a Western manufacturer's network.
The tool progressed through the Topcoder and Wazoku "Innovation Builders — AI Disruption" challenge, winning Stages 1 and 2 and reaching the Stage 3 vendor showcase. The working system comprises a live risk dashboard, an event feed and a manual simulation environment, exercised against a synthetic automotive dataset.
PRISM is a data model and enforcement layer for organisations that must know not only what their data says but where every statement came from and how well grounded it is. Built on the W3C PROV-O and Web Annotation standards with SHACL enforcement, it scores material along two independent axes — assurance of the source and grounding of the claim — so that downstream reasoning can be weighted accordingly.
PRISM was shortlisted to the final three in a UK government intelligence and security challenge, and the evaluators' feedback has been folded back into the design. It represents the practice's answer to a question that generative AI has made urgent: how an organisation reasons over material it did not author and cannot fully trust.
CascadeWatch asks whether the destabilisation of the Atlantic overturning circulation can be detected in advance from observational sea-surface data, using a network statistic — the leading eigenvalue of a lag-aware correlation matrix across physically distinct ocean regions — rather than the single-site indicators that dominate the literature.
The project is run to a standard unusual outside academia. Every analysis is pre-registered before data is touched, positive and negative results are committed in advance, and model-world tests that failed have been reported as plainly as the observational analysis that succeeded. The work is conducted with informal advisory input from Professor Tim Lenton's tipping-points group at Exeter, where it is the subject of an invited seminar.
The hardest question in any eradication campaign is not how to remove the target but how to know, defensibly, that removal is complete. Absence of detection is not detection of absence, and campaigns that cannot tell the difference either stop too early or pay for suppression indefinitely.
This engine addresses the problem as an inference task. A stage-structured population model is combined with multi-modal detection data in a provenance-aware framework, generalising published inference-of-absence methods, and the output is a running, quantified confidence that eradication has been achieved — distinguishing a genuine eradication trajectory from permanent suppression. The engine was developed against the invasive brown tree snake problem on Guam, validated in simulation against published field data, and offered to US government stakeholders as a licensable tool with training and knowledge transfer.
Security claims about physical systems — that a device is unclonable, that tampering is evident, that a measurement is fresh — are usually made in prose and evaluated by intuition. This work develops a substrate-independent formal grammar of physical trust primitives with an accompanying checker, allowing such claims to be composed and machine-verified rather than asserted. The work was developed in engagement with ARIA's "Trust Everything, Everywhere" opportunity space.
Alongside the named tools, Goodwood House works within UK government innovation programmes on problems including the visual comprehension of complex sensor estates and thermal management for compact electronics, and has taken proposals through HMGCC Co-Creation and ARIA processes. The details of live programme work are not published here.
The inference practice sits on a broad engineering foundation, and Goodwood House maintains a deliberate sideline in open-innovation challenges across the physical sciences — partly because the discipline of the format, a real problem, a hard deadline and expert evaluation, keeps the practice honest. Some of these won, some were shortlisted, and some were closed when the evidence said stop. All of them are reported here the same way they were reported to the evaluators.
Goodwood House Ltd is the independent research and engineering practice of Michael Eccleston, a Cambridge-trained engineer whose prior career reached SVP and CIO level in global freight logistics. The company operates as a specialist subcontractor and consultant, taking problems from concept through to validated demonstrator.
The practice has a sustained record on open innovation platforms including Wazoku, InnoCentive and Topcoder, with prize-winning and shortlisted submissions across intelligence, environmental, pharmaceutical and industrial domains.
We take on a small number of engagements at a time, typically scoped as a feasibility study, a demonstrator build, or a licensed application of an existing engine with knowledge transfer.
hello@goodwoodhouse.com