Expert time is scarce
Assigning signals by hand takes years of experience and hours of attention for every compound.
NMR structure elucidation platform
From NMR spectrum
to molecular structure.
Upload a spectrum and let our bespoke machine learning algorithms propose candidate structures. Refine them in an interactive workspace, check the predicted ¹³C spectrum against your data, and do it together with your team. All the computing runs on our side.
The problem
Structure elucidation shouldn’t be a bottleneck.
Tools are scattered
Predictors, sketchers and 3D viewers live in separate programs, each with its own licence and installation.
Compute is expensive
Running ML models or simulations takes hardware and administration that most labs simply don’t have.
NMR is hard to learn
Students learn to read spectra from static textbook figures instead of working with real, interactive data.
How it works
One workflow, from raw data to a verified structure.
Upload & clean the spectrum
Drop a JCAMP-DX file, including Bruker variants. Set the spectrometer frequency, tune the baseline, mark solvent peaks and let automatic peak detection do the rest.
Machine learning algorithms propose candidate structures
Our models read the peak list and propose candidate structures. Each one is built from chemical groups you can inspect, so you never get a black-box answer.
Refine it your way
Edit at the level of chemical groups or individual atoms. The two views stay in sync, and a library of 374 named groups labels every piece of the molecule.
Verify against the experiment
Predict the ¹³C spectrum of your candidate and overlay it on the measured one. Compare candidates side by side and keep the structure that fits.
The workspace
Built around the way chemists think.
Groups, atoms, spectra and 3D in one place. Every edit is reflected everywhere at once.
Atoms and groups, in sync
Draw or edit atoms and bonds directly, and the molecule is decomposed back into named chemical groups automatically.
374 named chemical groups
Search by name, alias or SMILES, or browse by category, from methyl to morpholinyl.
Start from a name
Type “paracet…”, pick paracetamol and it lands on the canvas. You can also paste SMILES or InChI. The lookup runs on our own index, so your queries never leave the platform.
Interactive 3D
Rotate and inspect every candidate structure in 3D, right next to its spectrum.
Split structures your way
Choose how a molecule is cut into groups, from retrosynthetic rules to the fragments a spectroscopist would pick.
BRICSErtlKnown fragmentsMolCut
A workbench, not a form
- Dockable panels with View, Edit and Advanced layouts
- Your spectra history, ready to pick up where you left off
- Your own custom fragments, saved to your account
- A built-in guided tour for first-time users
Collaboration
Work on the same structure, at the same time.
Share a workspace the way you share a spreadsheet. Everyone sees the same spectrum and the same candidate, and every edit appears live.
- Live presence. See who is in the workspace and where they are working.
- Supervisor and student on one spectrum, instead of screenshots sent by email.
- CRO teams reviewing a structure together, and with their client.
Anna
Marek
Prof. Kim
Learning mode
Learn NMR by doing, not by memorising.
Guided exercises on real spectra teach you to read shifts, recognise groups and reason your way to a structure, with feedback at every step.
- Students: practise on the same tools researchers use.
- Educators: run hands-on classes without installing anything.
- Anyone curious: learn chemistry and NMR at your own pace.
Try it · Carbonyl carbons
1 / 4
Which peak belongs to the C=O carbon?
Compound: naproxen
Pick a peak on the spectrum or a value below.
Compute included
All the heavy lifting runs on our side.
No GPUs, no installation, no software licences. Open a browser, and the machine learning algorithms, NMR prediction and soon molecular simulations run on MoleculeSolver’s own hardware.
Your browser
Upload, edit, compare
MoleculeSolver infrastructure
Machine learning algorithms
¹³C NMR prediction
Simulations · soon
Nothing to install
Everything runs in the browser: no installers, no licence servers, no waiting on IT.
Our own hardware
Models and simulations run on machines we operate ourselves, not on a patchwork of third-party services.
Your laptop stays free
Heavy computation happens on our servers, so even a student laptop is all you need.
Coming soon
From structure to biology.
Once your structure and its spectrum agree, take it further in the same workspace, on the same compute.
Force fields
Molecular mechanics and dynamics for your candidate structure, set up in a few clicks.
Membrane interaction
Simulate how your molecule behaves in contact with a complete cell membrane.
Ligand–protein binding
Check how a candidate ligand binds its protein target before you ever reach the bench.
Partners
Team
Chemistry, software and machine learning under one roof.
Rafał Stojałowski
CEO & Co-Founder
Trained in biotechnology and computer science and working professionally as an automation developer, Rafał connects laboratory practice with software engineering and leads the company’s strategy.
Michał
CTO & Co-Founder
A solution architect with many years of industry experience, Michał is responsible for the platform’s architecture, scalability and technical roadmap.
Jan Grubba
CRO & Co-Founder
Holding a master’s degree in machine learning, Jan leads research into new methods for structure elucidation and turns them into the models behind the platform.
Albert Masiak
COO & Co-Founder
Holding a master’s degree in artificial intelligence, Albert develops the platform’s machine learning models and oversees day-to-day operations and delivery.