Oil and gas
SciLM Subsurface
The first language model to see and generate 3D geology, seismic, and oil production to do both history matching and field development planning.
Read moreThe AI model
that can see physics.
We provide SciLM, a multimodal language model that can input and output any type of scientific and engineering data in its native form.
Stop describing your data to LLMs. Show it to SciLM.
Motion is reduced on this site.
Now, two pretrained specialized models.
Next, SciLM for all of science and engineering.
Use out of the box. No need to train on your data.
Our servers or yours. Your data can stay in-house.
Chemicals and catalysis
SciLM Atoms
A language model that sees and generates 3D atomic structures to propose a solution to your catalyst deactivation.
Read more3D meshflows and fields
-
Oil and gas
Darcy flow
CO2 storage
Darcy flow
Heat transfer
heat equation
Aerodynamics
Navier-Stokes
Chip design
Maxwell and heat
Climate
Navier-Stokes and radiation
Points in 3D spaceatoms, particles, and point cloud
-
Catalysts
surface reactions
Semiconductors
diffusion and growth
Batteries
ion transport
Pharma
molecular binding
Granular flows
particle collisions
LiDAR
3D sensing
One SciLM
for all of science
and engineering
The two highlighted tiles are the products shipping in the first half of 2027. Every other tile is the same approach in other domains.
Our domain expertise and data engines feed the models.
First
Simulate.
Experiments are too few and inconsistent. So we use physics-based simulations to generate training data at scale.
32 M reactions · 1 M reservoirs
Now
Pretrain.
Specialized models that work on your data out of the box.
SciLM Subsurface · SciLM Atoms
Next
Unify.
One SciLM for all of science and engineering. Run any task end to end.
You skip the training. Start on day one.
NeurIPS 2026
Two accepted papers present our data engines, datasets, and models.
What SciLM changes in a scientific discovery or engineering design loop.
Lower cost
SciLM gives better guesses that lead to fewer simulations, experiments, and costly mistakes.
Better solutions
Faster built-in surrogates of physics let you try more versions of your plan to reach the optimal solution.
New capabilities
Multimodality lets you consider all specialists in one pass to produce results considering all constraints.
True automation
Multimodality removes reliance on manual visualization and analysis of 3D/4D data inside the loop.
News.
Founded by two computational scientists/engineers.
Wrote a PhD dissertation at UT Austin on generating the training data AI in science needs. Created SaddleMill and ResMill, the engines behind both SciLM datasets, and SiliciclasticReservoirs, one million 3D reservoir models. Trained SaddleFlow, the first generative model for transition states in inorganic materials, and ResFlow, the first single model for every siliciclastic reservoir type, any wells, any field size. Built PotMill at Los Alamos National Lab, which automates ML interatomic potentials from data generation to the final model in under one day. Petroleum engineer by training.
Ilgar Baghishov
Co-founder and CEO
Advances computational science through machine learning, AI, and agentic workflows, with a focus on extending atomistic simulations to experimental timescales, as part of his PhD at UT Austin. Generated MaterialsSaddles (NeurIPS 2026), the first general-purpose open dataset of materials transition states (34M, ~40,000x larger than the closest prior). Developed an ML interatomic potential using 223x less memory than UF3. Architected and administers three HPC clusters (150 nodes). Biomedical engineer by training, with US and European patents pending.
Sung Hoon Jung
Co-founder
Scientific advisors
Developed the nudged elastic band and dimer methods and the Bader charge analysis code: the standard ways to find transition states and charges in materials simulation. Cited more than 100,000 times.
Graeme Henkelman
Professor of Chemistry, Oden Institute for Computational Engineering and Sciences, UT Austin
Spent thirteen years with Chevron’s Energy Technology Company before joining UT Austin, modeling the subsurface and conducting and leading research in energy data science and AI. Authored the textbook Geostatistical Reservoir Modeling (Oxford), the e-books Applied Geostatistics in Python and Applied Machine Learning in Python, and created the GeostatsPy Python package. As the GeostatsGuy on YouTube, teaches geostatistics and machine learning to tens of thousands of learners worldwide.
Michael Pyrcz
Professor of Petroleum and Geosystems Engineering, Jackson School of Geosciences and Oden Institute for Computational Engineering and Sciences, UT Austin
Spent seven years at Sandia National Laboratories as a core developer of Peridigm, the massively parallel code for simulating how materials fracture. Wrote the open course books on numerical methods and high-performance computing his UT Austin students learn from, and co-founded daytum.
John T. Foster
Professor of Petroleum and Geosystems Engineering, Oden Institute for Computational Engineering and Sciences, UT Austin