Image credit: immunitoAI — official company visual.
Finding an antibody that locks onto the right spot of a disease-linked protein is one of drug discovery’s slowest jobs. Done the conventional way, going from an idea to a lab-ready antibody candidate takes four to five years — most of it spent screening animals, human donors or synthetic libraries for a molecule that happens to bind. ImmunitoAI, a Bengaluru startup, wants to skip the search: its AbInitio platform designs antibody drugs computationally, from scratch, aimed at any protein target the company picks.
A visualisation of the molecular world immunitoAI works in — antibodies designed against disease-linked protein targets. Image: immunitoAI.
Biology first, then the machine draws
ImmunitoAI describes its philosophy in three words: “Biology First, Drug First, Intelligence Driven.” The process starts not with the algorithm but with the disease: the company’s biologics team defines the epitope — the precise spot on the target protein the antibody must grip — and the mechanism of action, whether the antibody should block or activate its target, before a single molecule is generated. AbInitio then designs directly against that epitope, generating novel antibody sequences de novo rather than screening a pool of existing candidates.
The platform’s neural networks train on 3D structures of binding protein pairs from the Protein Data Bank, a public repository. A mathematical transformation developed by the startup converts the atoms’ 3D coordinates into two graphs — atoms as points, interactions as links — so the model learns how two proteins’ shapes fit together, grooves accommodating bumps, and which amino acids bind to each other. Because the platform assesses drug-like properties — stability, affinity, specificity — at the design stage, the company says it can reduce the rework that usually comes later. Its claims, attributed: the AI-led process compresses the traditional 4–5 year timeline to 11–12 months, and CEO Aridni Shah has said the company reached functional candidates in less than five months from a single design step.
More Read
From the computer to the wet lab
AbInitio’s models learn the structural complementarity between antibody and target. Image: immunitoAI.
“We take roughly 1,000 to 10,000 designs to the lab, depending on the strategy,” CTO Trisha Chatterjee told YourStory. “An AI platform can generate binders, but which of them really bind in the real world is what matters.” The designs move into immunitoAI’s in-house biological lab, where they are characterised for target binding, specificity and drug properties. Experimental data then feeds back into the company’s proprietary dataset, retraining the neural networks — a loop the company says steadily improves the models.
The startup is currently pre-revenue and its candidates remain at the preclinical stage — the honest caveat behind the ambition. The company says candidates tested in cell-based assays have shown functional relevance and potency, and the next milestone is animal data, which it hopes will help close licensing deals with pharmaceutical companies.
Four programmes, two ways to sell
ImmunitoAI says it has four in-house antibody programmes in lab validation — each first-in-class or best-in-class in its telling — spanning oncology, autoimmune disorders, haematology and neurology. Programmes reported so far include CD40, GITR and CXCR4. The business model is platform-plus-asset: out-licensing its own drug candidates, or taking a pharma partner’s chosen target and identifying the final candidate for them.
The company was founded in November 2020 by Dr Aridni Shah (CEO), who holds a PhD in biological sciences from NCBS-TIFR, and Trisha Chatterjee (CTO), a computer-science postgraduate who previously worked in R&D at GreyOrange. The two met in August 2020 through Entrepreneur First’s Bengaluru cohort — Shah brought the biology, Chatterjee the machine learning. In November 2025 the startup raised a $6.1 million (Rs 52.14 crore) Series A led by investor Ashish Kacholia, with participation from 3one4 Capital and existing backers pi Ventures and Anicut Capital, taking total funding to Rs 66.98 crore. Earlier in October 2026 it was named to YourStory’s Tech30 cohort of promising Indian startups.
immunitoAI’s product demo. Video: immunitoAI (official company channel).
What to watch
The question for every AI drug-discovery company is the same: do the designed molecules hold up in biology? ImmunitoAI’s animal data, expected ahead of any licensing deal, will be the first real test. Then comes the commercial question — whether a preclinical antibody deal can run to the triple-digit millions the company projects — and the platform question, whether AbInitio’s next upgrade really means “very few candidates” need lab testing, as Chatterjee hopes. With plans to expand into the US and Europe, this is Bengaluru’s TechBio bet moving from code to clinic, one epitope at a time.
