The memory of spices, for me, begins with the click of a steel masala dabba opening in my mother’s kitchen. She would press her fingers into chilli powder to test its sharpness, crush coriander between her palms, and adjust proportions instinctively. Indian spice blending has always functioned as a living sensory system. It responds to humidity, to the age of stored turmeric, and to whether the dish must comfort or invigorate. A tadka changes slightly in the monsoon. A winter curry tolerates more warmth. These decisions are chemical in effect, even if never articulated in chemical terms.
Now, this unspoken chemistry is being translated into code. Across research laboratories and commercial kitchens, artificial intelligence is being trained to analyse thousands of volatile flavour compounds. By mapping shared molecular structures between ingredients, machine learning models can predict statistically compatible pairings, even when those combinations appear culturally unfamiliar. Long pepper with dark chocolate. Nutmeg with citrus peel. Fenugreek greens with fermented chilli.
A global phenomenon
Professor Ganesh Bagler of IIIT Delhi has spent years studying flavour networks and ingredient compatibility, positioning food within the language of data science.
“AI-driven spice blending represents a shift from trial-and-error to evidence-guided creativity,” he explains, and adds, “By analysing flavour molecules and historical culinary data, algorithms can suggest unconventional yet harmonious pairings.”
Databases that catalogue flavour molecules now allow ingredient chemistry to be computed and compared at scale. Globally, food companies are using AI tools to shorten product development cycles, reduce failed prototypes and optimise formulations before they reach physical kitchens. Instead of testing dozens of variations manually, developers can filter possibilities digitally, narrowing the field before real-world trials begin. India’s growing flavour and seasoning market, expanding alongside processed foods, quick-service restaurants and premium dining experiences, is beginning to explore similar approaches.
Craft under heat
At Farzi Beach in Goa, Executive Chef Omkar Gothankar encounters AI not in a laboratory, but in the immediacy of the stove. “I treat AI like a flavour rebel in my kitchen,” he says. “It dares me to try combinations that would sound heretical in a traditional spice box.” For him, however, suggestion is only the beginning. “Flavour in Indian cuisine must endure fire, smoke, fermentation and time, not just data.”
The distinction is crucial. A pairing that appears harmonious on a molecular map must still withstand hot oil spluttering in a kadhai, the slow caramelisation of onions, and the way spices mellow after prolonged cooking. Technique disciplines innovation. The chef does not surrender authority to the algorithm; he tests it.
Scaling flavour
Where chefs see creative provocation, technologists see operational possibility. Ankush Sabharwal, Founder and CEO of a conversational and generative AI platform, believes the real transformation lies not merely in experimentation but in execution and personalisation at scale.
“AI is fundamentally re-engineering the way we think about food personalisation and flavour innovation,” he says. “We are entering an era where culinary science meets machine intelligence, where algorithms can analyse thousands of flavour compounds, cultural food patterns and individual taste preferences to create hyper-personalised spice blends.”
According to him, the system does not simply suggest combinations; it calibrates them. “The AI system can ingest, for example, 10 different spices and, based on the food or meal type and the user’s preferences, automatically select the specific spices and the optimal quantity for each,” he explains. That precision, he argues, shifts flavour design from instinct-driven approximation to measurable formulation.
Integrated with IoT-enabled dispensers and digital ordering systems, such platforms could allow restaurants to retain customers’ flavour preferences and reproduce them consistently. “By blending tradition with intelligent automation, restaurants can retain customers’ historical taste profiles and deliver food that matches personal preferences every single time,” Sabharwal says. He describes this as the rise of “physical AI” in food service and adds that it is a matter of implementation. In this model, AI becomes a memory engine, ensuring repeatability with variation.
On Goa’s coast, Puru, restaurateur and founder of Ourem, views AI through the lens of geography and cultural exchange.
“India’s food knowledge has historically travelled slowly, through migration, trade routes and kitchens,” he says. “Algorithms collapse that distance.”
A coastal flavour structure, built around acidity, coconut, fermentation and sea-influenced spices, can now be digitally mapped against Southeast Asian or East African systems almost instantly. What once required years of lived experience can be recognised in seconds through pattern analysis.
For Puru, the question is not whether AI replaces instinct, but how rapidly it accelerates cross-cultural discovery. Yet speed, he implies, does not automatically confer depth. A spice still responds to climate, oil, timing and touch, variables embedded in lived cooking rather than datasets.
Psychology of taste
Rohit Dadlani, founder of Pause Mumbai Cafe, shifts the focus from chemistry to perception.
“Taste is shaped by memory and nostalgia,” he says. “Two people can eat the same dish and experience it differently.”
His point introduces a quieter complexity. While AI may map molecular compatibility, it cannot easily account for childhood associations, regional identity or emotional memory. Flavour is not only about compounds interacting on the tongue; it is about recognition, comfort and familiarity.
Cyril Feuillebois, founder of a spa that uses Indian spices, approaches the development from a market perspective. AI-generated blends, he says, represent “an exciting and flavour-forward evolution,” one that allows brands to experiment more efficiently while expanding the boundaries of what consumers might accept.
Perhaps that is the clearest way to understand this moment. Indian cuisine has never been static. Chillies arrived from the Americas and became indispensable. Tomatoes were once foreign. Techniques crossed oceans and were absorbed into daily cooking. Each innovation survived only if it adapted to the logic of the Indian kitchen.
Artificial intelligence is now undergoing the same test. It may speak in molecules and models. But like every ingredient before it, it must prove itself in hot oil, in patient simmering, in the human memory of taste.
And somewhere, in kitchens across the country, Indian mothers will still open their masala dabbas, taste the salt, adjust the spice by instinct, and quietly decide whether the algorithm truly understands flavour.
Mix and match
Turmeric + star anise: Earthy warmth balanced with subtle sweetness; works well in slow-cooked gravies and broths.
Black pepper + jaggery powder: A bold sweet-spicy contrast ideal for chutneys, marinades, and glazes.
Coriander seeds + cocoa powder: Adds a warm, nutty depth to rich gravies and meat or vegetable curries.
Fennel seeds + smoked paprika: Creates a sweet-smoky flavour that pairs well with roasted vegetables.
Cumin + cinnamon: Brings comforting savoury warmth to lentils, rice dishes, and stews.
Fenugreek leaves + lemon zest: A bright combination with slight bitterness that lifts vegetable dishes.
Sesame seeds + black cardamom: Deep, smoky nuttiness suited for robust gravies and rice preparations.
Nutmeg + chilli flakes: A balance of gentle heat and warm sweetness.
Mustard seeds + orange zest: Sharp, citrusy notes that add freshness to tempering and marinades.
Cloves + honey: Sweet and aromatic, ideal for glazes or slow-cooked dishes.