Aridni Shah wanted to see where a few genes switch on inside a honeybee brain. The tool for the job was an antibody, and nobody sold one for honeybees.
That gap became Immunito AI. Aridni is co-founder and CEO of the Bengaluru company, which uses generative AI to design antibody drugs from scratch. In November 2025 it raised a $6.1 million Series A led by investor Ashish Kacholia.
Check out the video of the conversation here or read on for insights.
How did a honeybee PhD lead to an antibody company?
Aridni grew up in Darjeeling and moved to Delhi at 15 to study medicine. She missed the PMT entrance exam, took a B.Sc. in Biomedical Science instead, and went on to an integrated PhD at NCBS in Bengaluru, studying time memory in honeybees.
Antibodies are proteins that lock onto a specific target, and researchers use them to find proteins in tissue. Honeybees have no commercial antibody catalog, so a custom one was the only route. That meant a year to a year and a half of work with no guarantee it would bind.
“So imagine you have a five-year PhD, you spend a year, year and a half trying to make this antibody”
She switched to RNA-based visualization, but the bottleneck stayed with her. In 2020 she joined Entrepreneur First’s fourth Bengaluru cohort and met Trisha Chatterjee, who has a master’s in computer science with an AI focus and about 10 years of robotics experience at GreyOrange. Trisha is the company’s CTO. Aridni brought the biology and Trisha brought the machine learning.
The pharma version of the problem is bigger. The standard method is a form of vaccination: inject a target protein into mice or rabbits, wait three months to a year and a half, then screen 10 million to a billion antibody sequences down to 10 or 20 candidates. Those leads often clump or degrade at drug concentrations, so researchers mutate them and test again. Aridni says only about one in 50 drug projects reaches the market. The prize explains the persistence: Merck’s antibody cancer drug Keytruda sold $29.5 billion in 2024.
How does Immunito AI design antibodies with AI?
Immunito AI starts from the target’s 3D shape and builds an antibody to fit it, like cutting a key for a known lock.
“I am saying I know what my lock looks like. I want to just build the key for it.”
AlphaFold cannot do this, Aridni explains, because it needs a sequence before it can predict a structure. Immunito AI has only the target and must produce the sequence itself.
Data is scarce. The Protein Data Bank holds about 200,000 protein structures, and only 9,000 to 10,000 are antibody-antigen complexes. So the team encodes each atom and its neighbors as a graph, then trains a diffusion model, the method behind many image generators, to design a matching antibody.
The output is an amino acid sequence, which converts directly to DNA. Generation takes hours to days. Synthesis takes about a week for a few sequences and six to eight weeks for a library of 100,000 or more. The company says this cuts a 4 to 5 year discovery timeline to 11 to 12 months. Filters for stickiness and instability run during design, and the in-house wet lab tests every candidate.
“You have no idea what the animal has given you and you’re basically trying to work with that”
How is Immunito AI funded, and how will it make money?
The company raised a $1 million seed led by pi Ventures in September 2021, then the $6.1 million Series A, for more than $7 million in total. Aridni says Kacholia put in about $3 million, pi Ventures $1.25 million and Anicut Capital $1 million. 3one4 Capital and AC Ventures joined as new investors.
Inc42’s cap table data from November 2025 shows founders holding 35.3%, with Trisha the largest single holder at 17.7%. Investors hold 50.2% and the ESOP pool is 14%. The team is about 25 people, and FY25 revenue was ₹5.8 lakh.
Aridni credits one investor’s conviction for getting the round done.
“The gut of Ashish Kacholia, I should say, one person believing in a lot of deep tech companies in India.”
Other funds were harder. She says some talks ran six months before ending without a decision.
“They don’t have domain experts to actually kind of, you know, be able to review us and exactly underwrite.”
The contrast with the US is sharp. Chai Discovery, a San Francisco rival also designing antibodies with AI, has raised more than $225 million and is valued at $1.3 billion. Aridni says the $6 million lets Immunito AI do validation work “we should probably have done a year or two ago.”
The model is platform plus asset. Four in-house targets in cancer and autoimmune disease are in lab validation, each first-in-class or best-in-class. Pharma partners can bring their own target, but they get lab-validated drug candidates, not access to the model. Aridni says preclinical antibody deals typically run to triple-digit millions overall, with $10 million to $50 million upfront, then milestone payments and royalties. The buyer takes the IP and the trial costs, which is why Immunito AI does not plan to build a sales force. The next milestone is animal data, which she expects to help close those deals.
What does it say about Indian biotech?
India’s pharma industry grew on generics and biosimilars, with Biocon and Enzene making copies and Syngene running outsourced development. Novel antibodies in human trials are rare. Aridni points to Zumutor Biologics, which is headquartered in Boston with R&D in Bangalore and began a Phase 1 trial of a first-in-class antibody in 2024.
Policy is shifting. The BioE3 policy, approved in August 2024, targets a $300 billion bioeconomy by 2030. Aridni’s concern is the size and speed of grants. She notes that a European competitor received a €2 million grant, something she says is hard to imagine in India today.
Her argument for backing discovery is short.
“Services, you can do a certain extent, but the true value is in the IP itself”
Listen now!
Other ways to listen:
Sources