Simulate molecules no databank has
ML and physics predict thermodynamic and transport properties from structure, anchored to measured data wherever it exists.
Process simulation for molecules that have never been made
Know whether a new molecule will run in your plant, before you build a pilot.
Morio predicts the properties of molecules no databank has seen, puts a calibrated error bar on every one, and runs them through your flowsheet to purity, energy and cost. You see the answer, and how far to trust it, before anyone commits capital.
- Aspen Plus
- Aspen HYSYS
- DWSIM
- gPROMS
Purity · overhead99.4 % ±0.3
Reboiler duty2.1 MW ±0.2
Sweep · overnight2,048 designs
The problem
Your simulator has never met your molecule.
AI and high-throughput labs now propose candidates faster than any team can evaluate them. Yet only about a hundred new chemicals reach US production each year, because nobody builds a plant around a molecule they cannot simulate.
No data, so somebody guesses.
A new compound is in no databank. Its properties get estimated by hand, with group-contribution methods that are decades old.
One number, no error bar.
A guessed viscosity looks exactly as certain as a measured one. Nobody can say how far the purity or the energy bill could move.
A handful of candidates, not hundreds.
Every flowsheet is built by a scarce process engineer, so most of the molecules your chemists or your models generate are never looked at.
The pilot plant becomes the experiment.
Years and capital go into finding out what the simulation could not tell you. The surprises arrive when they are most expensive.
What Morio does
Morio makes a never-made molecule simulation-ready, and tells you how far to trust the result.
Know how wrong the answer could be
Every property carries a calibrated band, and the band follows it through the flowsheet to purity, energy and cost.
Screen every candidate, not three
Agents build and sweep flowsheet variants for each candidate and rank them on what matters at plant scale.
Keep the simulator you trust
Export a CAPE-OPEN property package into Aspen Plus, HYSYS, DWSIM or gPROMS. Your validated models stay where they are.
How it works
Error bars from the molecule to the plant.
Other tools stop at the property, or hand you a single number. Morio carries the uncertainty all the way through, so the risk is on the table before anyone builds anything.
Molecule
SMILES · structure · nameBring a structure. No measured data needed.
Properties
Viscosity 38 mPa·s±6Predicted from structure, with a calibrated band and a source on every value.
Flowsheet
Reboiler 2.1 MW±0.2Agents run your process with the bands attached and sweep the design.
Decision
Purity 99.4 %±0.3Purity, energy and cost with error bars. Export to your simulator, or make the call.
The product
Every property carries a band and a provenance, so your engineers see what is measured and what is predicted.
Property package · IL-7 · new compoundFive of eight properties are predicted; the viscosity band drives most of the duty uncertainty.
Band on reboiler duty±9 %carried from the property bands
Measured · Predicted3 · 5ILThermo points found for 3 of 8
Widest band±16 % viscosityone measurement at 40 °C would halve it
PropertyValue95 % bandSourceStatus
Vapour pressure · 80 °C12.4 kPa±1.1Predicted · GNN + COSMO-RSTolerance
Viscosity · 40 °C38 mPa·s±6Predicted · GNN · widest bandTolerance
Heat of absorption · CO₂−58 kJ/mol±4Measured · ILThermo · 6 pointsConverged
Density · 25 °C1.08 g/cm³±0.01Measured · ILThermo · 14 pointsConverged
NRTL τ₁₂ · CO₂ in IL-70.42±0.09Fitted · 11 predicted VLE pointsTolerance
Show calibration curvesExport CAPE-OPEN package
Provenance on every number
Each value shows whether it was measured, predicted or fitted, and from what.
Proposals your engineers approve
The agent suggests a change and shows the band it moves. Nothing changes until an engineer says so.
The next measurement that matters
Morio tells you which single lab measurement would shrink the uncertainty most, so the lab works on what moves the decision.
Works with your stack
Morio works with the simulator you already run.
Results export through CAPE-OPEN, the standard way to plug a property package into a process simulator. Your plants keep their tools and their validated models. Morio adds the part none of them have: properties you can trust for a molecule no databank has seen.
What changes for a new molecule
| Today | With Morio | |
|---|---|---|
| Properties | Hand estimates from decades-old group-contribution methods | Predicted from structure by ML and physics, anchored to measured data |
| Uncertainty | One number, no error bar | A calibrated band on every property |
| Plant-scale impact | Unknown until the pilot plant | Bands carried through to purity, energy and cost |
| Candidates evaluated | A handful, one flowsheet at a time | Every candidate, swept by agents |
| Workflow | Desktop tools, scripted by hand | API-first and agent-native, with an MCP server |
Who it is for
Built for the teams that have to bet on a new molecule.
R&D and innovation leaders
Make the go or no-go call on a new molecule before the pilot plant, with the risk on the table.
AI discovery teams
Rank candidates by whether they can be made and run at scale, not only by how they perform in the flask.
Process engineers
Get a property package with error bars for compounds no databank has, inside the simulator you already use.
- Carbon capture
- Specialty chemicals
- Solvents and process fluids
- Polymers and elastomers
- Semiconductor materials
- Energy and refining
Case study · Refinery CO₂ capture
We have taken a designed molecule all the way to a refinery digital twin.
Our team designed a CO₂-capture molecule, had it validated independently by URV in Spain, and supplied its parameters to SOCAR's Aspen Plus digital twin of the STAR refinery, where it captured 67 % of CO₂ at 15 bar. Morio is that loop, turned into software.
- 67 %
- CO₂ captured at 15 bar in SOCAR's refinery twin
- 8
- candidates synthesised and tested in three months
- URV
- independent validation of the designed molecule
Pilots
Bring us a molecule nobody has made yet.
Tell us about the candidate and the process it has to run in. We will scope a pilot on your own flowsheet.
- 01Tell us about your moleculeOne email: the candidate, the process, the decision.
- 02Thirty minutes with our foundersWe go through your flowsheet and what you need to know.
- 03A scoped pilotOne molecule, your process, a clear answer with error bars.