StratoBayes uses Bayesian AI to automate subsurface correlation, quantify geological uncertainty and help geoscientists make faster, more defensible decisions across energy, carbon storage, mining and infrastructure.
Durham University spin-outPeer-reviewed methodPatent published
Subsurface interpretation today is manual, slow, subjective and hard to defend. One in five wells misses its target, and a mis-placed well incurs costs in the millions of pounds. StratoBayes produces ranked interpretations with quantified uncertainty, helping geologists and exploration teams reach better decisions in hours instead of weeks.
Manual correlation of multi-well datasets can take weeks. StratoBayes delivers ranked correlations in hours, so geologists spend their time on interpretation rather than alignment.
Every result carries quantified probabilities and alternative scenarios, so you know how much weight each correlated horizon can bear.
Output is reproducible and auditable, built for regulatory evidence and investment cases and grounded in peer-reviewed mathematics instead of black-box predictions.
StratoBayes evaluates your data against a peer-reviewed statistical model, weighs competing correlation hypotheses and states how confident it is in each one.
Import depth-referenced records from two or more boreholes, wells or outcrop sections, such as well logs, geochemistry and lithology.
The Bayesian engine evaluates thousands of candidate alignments, allowing sedimentation rates to vary between sites and updating its probabilities as the evidence adds up.
You receive the most probable correlations, ranked, with uncertainty estimates for every matched horizon, ready to inspect, adjust and export.
Two wells in, matched horizons out, uncertainty quantified at every step.
Set up the run yourself, or tell Strata, the built-in correlation assistant, what you want.
A live model run, recorded in the app.
The Strata assistant sets up runs, plots curves and reports diagnostics in plain language.
Correlate storage and monitoring wells with quantified uncertainty for site characterisation and the evidence regulators expect.
Trace target horizons between wells to estimate reservoir continuity before committing to the next borehole.
Align drillhole geochemistry across a deposit to follow ore-bearing horizons and plan follow-up drilling.
Automate multi-well log correlation to keep reservoir models current through appraisal and development.
Build a probabilistic picture of the ground between site-investigation boreholes before tunnelling or foundation design.
Correlate cores and outcrop sections on to a shared timescale using the method published in Geochronology.
StratoBayes is a Durham University spin-out, completed in June 2026, building on research licensed from the University.
The method is peer-reviewed: Eichenseer, K., Sinnesael, M., Smith, M.R. and Millard, A.R. (2025), StratoBayes: a Bayesian method for automated stratigraphic correlation and age modelling, Geochronology 7(4), 545-570 (read the paper). The underlying research was funded by the Leverhulme Trust.
The international patent application is published as WO 2025/196391 A1, with the UK patent expected to grant in September 2026.
StratoBayes Ltd is the commercial venture built on this research. Access to the software is through the beta programme.
"Dealing with correlation uncertainty addresses a persistent industry problem."
Hotspur Helium · structured trial partner
Today the engine correlates boreholes with quantified uncertainty. Next, we are building a foundation model for the subsurface, with the same Bayesian engine at its core, so machines can learn the statistical structure of the ground the way weather models have learned the atmosphere.
Sign up to join our beta programme and you'll receive an email with installation instructions.
Prefer a walkthrough first? Book a demo.
