Excelra makes
the past useful.

A molecule can have a long paper trail. It appears in a patent, turns up in an assay, acquires a measurement, then disappears into somebody else’s filing system. The result exists. Reusing it is another matter. Excelra has made that gap its business: finding scientific results, giving them structure, and delivering them to people deciding which experiment to run next.

This is an interesting place to stand in the AI economy. A drug-discovery model needs examples of what molecules have done. Those examples arrive with different terminology, formats and experimental contexts. Excelra’s wager is that organizing the record remains valuable even as the machinery reading it becomes more sophisticated.

A memory for molecules

Its flagship family, GOSTAR, takes its name from Global Online Structure Activity Relationship. The important word is relationship. A chemical drawing alone tells a researcher much less than a drawing connected to biological activity, safety observations and the conditions under which a result was obtained. The product gives those pieces a searchable home.

The range reflects different scientific questions. GOSTAR TPD focuses on targeted protein degraders, while Large Molecules offers tailored curation for biologic research. Beyond chemistry, Excelra builds custom biomarker knowledgebases. The value in each case depends on selecting evidence that fits the researcher’s question, rather than treating every scientific record as equally useful.

Excelra describes a process involving manual standardization and three levels of quality control by medicinal chemists. Customers can use a web application, obtain datasets or connect through an API. That choice matters: a scientist investigating one compound and a developer training a model need different ways into the same body of evidence.

The customer examples make the use concrete. Orogen Therapeutics announced in December 2024 that it would incorporate GOSTAR datasets into a discovery platform combining DNA-encoded libraries, machine learning and molecular simulations. Variational AI describes GOSTAR data as an ingredient in Enki, its generative chemistry technology. Fauna Bio uses Excelra to curate datasets spanning adapted animal species and human biology.

“Excellent AI discovery relies on clean, well-curated datasets.”Ashley Zehnder, CEO, Fauna Bio

There are alternatives. ChEMBL, maintained by EMBL-EBI, is an open, manually curated resource joining chemical, bioactivity and genomic information. Companies can also assemble their own datasets. Excelra’s commercial case rests on the usefulness of its coverage, quality controls, delivery and human support for a particular research problem. A larger collection alone cannot settle that buying decision.

A scientist who speaks software

Anandbir Singh Brar did not arrive through a conventional medicinal-chemistry career. He studied economics and international relations, earned an INSEAD MBA, and worked in energy finance and renewable energy. In a 2022 interview, he traced his involvement to the team identifying Excelra’s opportunity in 2014. He became CEO in 2016, the founding year the company now lists.

Brar’s description of the workforce is revealing: bilingual in science and technology. A programmer can move a measurement between databases. A scientist can ask whether two measurements belong together. A useful research system needs both judgments, sometimes from the same person. That combination explains why Excelra looks partly like a database publisher and partly like an engineering consultancy.

Its work extends into bioinformatics, clinical data, knowledge graphs, laboratory software and cloud infrastructure. The business model follows: subscriptions and platforms, consulting assignments, embedded specialists and longer partnerships. The company’s stated values include trust and teamwork; it also announced a third consecutive Great Place to Work certification in 2024.

The expansion was customer-led, according to Brar’s interview. Analytics developed into a broader data-science capability, while an internal research IT support function became a business unit. This suggests a practical explanation for the sprawling service menu: once data is organized, clients need help moving it, analyzing it and fitting it into daily work.

The ₹123.33 crore next step

In April 2023, Excelra bought BISC Global, adding bioinformatics consulting and platforms. BISC brought expertise in sequencing analysis, statistics and machine learning, alongside tools including the Online Pipeline Platform, or OP². Today, OP² lets researchers upload sequencing data, choose workflows and run analyses using cloud resources.

That price comes from ICRA’s October 2023 credit report. It also records a less celebratory detail: Excelra’s operating margin fell from 38.3% in FY2022 to 31.4% in FY2023 as lower-margin technology services contributed more and employee costs increased. Revenue still grew. Selling a broader solution can bring more work without preserving the economics of the original mix.

The same report put about 67% of revenue with five customers. Those are historical figures, but they explain a durable management question: how do you deepen valuable research relationships while widening the customer base? Adding capabilities creates more opportunities to sell; it also creates more obligations to deliver.

The next experiment still matters

GOSTAR’s July 2026 retrosynthesis release shows where the product is heading. Researchers can explore AI-assisted routes for making a compound alongside its scientific record, with checks on the commercial availability of proposed intermediates. The workflow moves from examining what a molecule did toward planning how to make it.

Excelra presents this as decision support before experimental validation. That boundary matters. A plausible route still needs a chemist; a promising prediction still needs an experiment. The September 2026 announcement of OpenAI Select Partner status extends the AI story, but the underlying test remains scientific usefulness.

That also changes how a buyer should think about cost. A database license is one part of the work; connecting it to internal records and maintaining a useful workflow are others. A team whose existing data already answers its question may have little reason to add another product. The purchase has to earn its place in the experiment.

For a prospective customer, the sensible starting point is a narrow question. Does the dataset cover the relevant target and assay? Can the team inspect provenance, preserve context and move results into its own systems? Start there. Excelra’s story offers a transferable idea: make yesterday’s result usable before asking tomorrow’s model to learn from it.