In silico cocrystal screening
Turn a poorly soluble candidate into a developable solid form.
CocrystalAI screens your active ingredient against hundreds of coformers in silico and returns the pairs most likely to form a cocrystal, ranked and ready for the bench. No material spent, no months of trial and error, no waiting for a vendor report.
The problem
70–90%
of drug candidates in development are poorly soluble.
Solubility is where good molecules die.
They show activity in the assay and then fail to dissolve, fail to absorb, or fail to stay stable long enough to become a product. The chemistry is right. The solid form is holding it back.
What a cocrystal unlocks
One partner molecule. A new solid form. A different drug product.
A cocrystal pairs your active ingredient with a pharmaceutically acceptable coformer inside one crystal. The molecule is untouched. The material is new, and with it come the properties that decide whether a candidate moves forward:
Higher solubility and faster dissolution
Better oral bioavailability for BCS class II and IV compounds
Improved physical and chemical stability
Lower moisture uptake and better handling
A form that can be compressed, tableted and scaled
A new solid form you can patent
The bottleneck
The right coformer is out there. Finding it by hand is the problem.
There are thousands of acceptable coformers. Every experimental attempt costs days of laboratory time and milligrams of an API you do not have to spare. So screening campaigns stop at a few dozen candidates, usually the familiar ones, and the coformer that would have worked is never tried.
The solution
Screen the whole library first. Crystallise only what deserves it.
Enter your active ingredient as a structure. Choose the coformers, from a handful to your full library. CocrystalAI evaluates every pair and returns a ranked shortlist of the candidates most likely to form a cocrystal, in minutes, without a milligram of material. Bench time goes to the pairs that earn it.
Shortlist · your active ingredient
8 of 8 screened
1
Nicotinamide
Likely
2
Isonicotinamide
Likely
3
Nicotinic acid
Worth trying
4
Succinic acid
Worth trying
5
Urea
Unlikely
Illustrative screen. Every row in the product carries its own evidence and record.
Why you can trust it
Every candidate comes with its reasons.
Confidence you can read
Each prediction states whether your chemistry is within the model's experience. Pairs outside it are flagged, never silently scored.
Chemistry you can check
The hydrogen-bond pairings and the known cocrystals that support each candidate are shown beside its rank.
A record you can defend
Every screen is stored with its full evidence, exportable to PDF for the review meeting and reproducible later.
In silico and at the bench
The prediction and the experiment, in one place.
Most screening tools stop at the screen. CocrystalAI is the software half of a solid-state laboratory: what the model ranks, our bench crystallises, characterises and scales. One team runs both halves, and what the bench learns goes back into the model.
LAB
The shortlist goes to the bench.
Experimental screening, X-ray diffraction and thermal analysis confirm which predicted pairs are real cocrystals.
SCALE
From milligrams to kilograms.
A promising pair is crystallised, characterised and scaled on the same bench, into material you can formulate.
DATA
A proprietary database.
Years of solid-form work, measured in our own laboratory, feed the model alongside the public literature. Our own high-performance computing runs it.
TEAM
Crystal engineers and data scientists.
The people who read the diffractogram sit with the people who train the model. Every prediction can be argued with chemistry.
Intellectual property
A new solid form is new intellectual property.
A cocrystal is a new material, patentable on its own. Found early, it protects a molecule for years beyond its original patent. Found on a marketed drug, it opens a new formulation and a new life cycle. Found by a generic developer, it is a way into a crowded form space. The earlier the screen, the more the form is worth.
When to use it
Built for the cases where salts do not solve it.
Your API is neutral or weakly ionizable and salt formation is not an option
The salt you have is hygroscopic, unstable or poorly soluble
You need a differentiated solid form for lifecycle or generic development
You want to de-risk solid form selection before committing material
What changes
What changes with CocrystalAI
TIME
Months to minutes.
A full coformer library screened before the first experiment.
MATERIAL
Material saved.
Early-stage API goes into candidates, not guesses.
BENCH
Fewer failed batches.
The bench works from a ranked list with reasons, not from a hunch.
OPTIONS
More forms, sooner.
More solid forms on the table means more options for solubility, stability and protection.
How it works
Three steps in. Eight analyses out. One bench to confirm it.
01
Add your active ingredient
By name, structure or drawing.
02
Add the coformers
A few, or a whole library.
03
Get the ranked shortlist
With every analysis beside each pair, and export it.
04
Confirm it at our bench
Send the shortlist to our laboratory. We crystallise, characterise and scale the hits. Optional, and the same team.
What CocrystalAI evaluates for each pair
Cocrystal formation likelihood
The core prediction, from the molecular structures of the two partners, ranked across the whole library.
Applicability domain
Whether your chemistry sits within the model's experience, so you know which predictions to act on.
Molecular complementarity
Shape, size, polarity and hydrogen-bonding capacity of each partner, and how they fit together.
Supramolecular synthons
The hydrogen-bond pairings the two molecules can form, the same motifs crystal engineers use to design cocrystals.
Solubility parameter match
Hansen solubility distance between the partners, an independent check on miscibility.
Similarity to known cocrystals
The closest reference cocrystals to your pair, so a prediction can be checked against precedent.
Literature support
Published work on the structures involved.
Explanation
Which atoms and groups drive the prediction, so the result can be argued with chemistry, not only with a score.
Each analysis is reported on its own. Nothing is blended into a single number you cannot question.
Who it is for
For preformulation, solid-state and CMC teams in pharma and biotech that need to move a candidate faster. For generic developers building differentiated products. For anyone who wants the screen done end to end, from the ranked shortlist to a confirmed solid form, by one team that owns both the model and the bench.
Where we are
Rio de Janeiro, São Paulo and the United Kingdom.
The laboratory is in Rio. The team works from Rio, São Paulo and London, and all three answer the same email.
Write to contact@cocrystalai.com or request a trial.
RIO DE JANEIRO
Laboratory and offices
Lead América Business
Av. das Américas, 2480, Bloco 2, Lojas 123 e 124
Barra da Tijuca, Rio de Janeiro, RJ, 22631-004
SÃO PAULO
Offices
Av. Eng. Luís Carlos Berrini, 1748
Itaim Bibi, São Paulo, SP
LONDON
International office
124 City Road
London, EC1V 2NX
United Kingdom
Give your best molecule a solid form that works.
Get in touch
Try CocrystalAI on your own molecule.
Request a trial, book a demo or ask about licensing. A few lines about your project is all we need to set up your first screen.
Reply within one business day. From the team, not an autoresponder.
Confidential by default. Your structures and results are never shared.
No card, no commitment. The trial is free and we scope it together.
Prefer email? Write to contact@cocrystalai.com.
A product of
CocrystalAI by Nanobusiness, InsilicAll and ConvergeLab
Research decision-support software.