AI for chemistry and materials R&D is an AI platform that helps chemists design new chemicals, materials and product recipes, from the first experiment to the final recipe. R&D means research and development. The final recipe of a product, such as a paint, an adhesive or a cream, is called its formulation. A formulation says which ingredients to use, in what amounts and in what order.
Why does a formulation need AI? A paint can contain 15 ingredients. Each one can be used in many amounts. So there are millions of possible recipes, and each one has to be mixed and tested by hand. Most companies have years of past experiments, but the results are spread across notebooks, spreadsheets and instrument files. Nobody can look at all of them at once.
The platform first collects that history in one place. Chemists record each new experiment in it, as an electronic lab notebook. The platform also receives test results from the instruments, through lab instrument integration, and sample records from the laboratory information management system (LIMS). Then a model learns from every past experiment how the ingredients affect the results. It predicts the results of recipes that nobody has tried yet.
Consider a specialty chemicals company in Vadodara that wants a new waterproof coating for roofs. The coating must dry within 4 hours and survive a monsoon. The company has records of 8,000 earlier coating experiments. The model is trained on them. The chemists ask for recipes that dry in under 4 hours, stay flexible at 45°C and cost less than ₹180 a litre. The model suggests 10 recipes and ranks them. The chemists mix and test those 10, instead of the 60 they would have tried by hand. Three meet every target. Their results are added to the records, and the next round of suggestions is better.
The same platform is used by drug companies. There, the formulation is the cream, tablet or injection that carries the medicine's molecule. The molecule's predicted properties can come from molecular modelling software. The platform then helps the chemists choose the other ingredients, so that the cream is stable and the molecule reaches the skin.
Some laboratories go one step further. They connect the platform to robots that mix and test the samples automatically. The model chooses each next experiment, and the robots run it, with no person needed between the steps. This is called a self-driving lab.
How is this different from AI drug discovery? AI drug discovery searches for one active molecule that acts on a disease. AI for chemistry and materials R&D works on the whole recipe, and on many industries, from paints and plastics to food and medicines. It is a part of computational drug discovery when the recipe is for a medicine.