Before anyone at Sorcero could make the product smarter, Dipanwita Das had to make the problem less vague. She interviewed roughly 300 people. Not 300 survey responses, not 300 clicks arranged into a dashboard, but hundreds of conversations with people who lived inside the work. She listened for the nouns they used, the decisions they dreaded, the information they could not reach and the steps a neat software demo tends to edit out. Only then did those conversations become product and technical requirements.
It is an origin story with a wonderfully inconvenient moral. The decisive technology was listening. Das, the CEO and co-founder of the Washington, DC company, did not begin by asking how much artificial intelligence could do. She began by asking where expert attention was being wasted, where scientific knowledge became stuck and what kind of system could be trusted in a consequential workplace. The machine came later, appropriately.
Approximate customer-discovery interviews Das conducted before Sorcero's early product took shape.
The problem arrived before the company
Das's route to enterprise AI ran through public policy and digital communication. At 42 Strategies, the agency she founded and led from 2014 to 2018, her work included digital transformation projects for public-interest organizations. Scientific communication had to travel between very different audiences: researchers, advocates, legislators and members of the public. The same evidence had to remain accurate while becoming usable in different rooms.
That experience left her with a persistent question. The world did not suffer from a shortage of documents. It suffered from the long, uncertain journey between finding knowledge and acting on it. Life-sciences organizations had an especially sharp version of the problem. They held large collections of publications, presentations, field reports and other unstructured material, while their experts still had to locate the relevant evidence, understand its context and explain it carefully.
At a colleague's dinner, Das met Walter Bender, the former MIT Media Lab executive who would become Sorcero's chief scientific officer. Their conversation moved through AI, personalized learning and the personalization of content. Richard Graves, an experienced entrepreneur whom Das already knew, joined as the commercial co-founder. The roles formed around their strengths: Bender on the technical thesis, Graves on bringing a product to market, and Das on the product vision that connected the customer's work to the possible system.
“Humans plus AI is greater than either alone.”Dipanwita Das
Sorcero launched in 2018, well before every strategic plan acquired a generative-AI paragraph. That timing gave the team no fashionable shorthand. Buyers were still asking whether they were purchasing AI, software, research or some ambiguous mixture of all three. The company had to develop the technology and educate the market while earning trust from early customers. Das has described the market as not fully formed, which is a courteous phrase for building the road while trying to sell tickets for the bus.
A vertical found by subtraction
Sorcero's initial thesis could have served several technical industries. Reality was less accommodating. A young company could not field a hundred specialists to solve small pieces of ten markets. Das and her co-founders saw that life sciences offered the clearest return for customers and the strongest match with her prior work. Focus arrived partly as strategy and partly as arithmetic.
The choice changed the product. Generic language tools can be impressive when the cost of a mistake is an awkward paragraph. Scientific work asks harder questions. Where did an answer come from? Does the system understand that two similar phrases mean different things in a particular discipline? Can an expert inspect, correct and validate the output? Sorcero combined language models with biomedical ontologies, traceable data and human review. Das's public shorthand has remained admirably unromantic: technology should augment the trained professional, not impersonate one.
The narrowing thesis
A conceptual view, not market-size data: each step trades breadth for domain context, traceability and workflow depth.
There is a useful discipline hiding in that architecture. Trust is not a legal paragraph at the bottom of a screen. It is a collection of product decisions: preserving provenance, exposing the evidence, inviting subject-matter review and constraining a system to the job it can reliably perform. Das has argued that AI in life sciences must be medically tuned and held to the rigor of the field. Her favored build equation places expert judgment inside the product, not outside it as an apology.
The reader in the engine room
For someone preoccupied with infrastructure, Das has a taste for imaginative worlds. She has spoken about reading The Lord of the Rings and The Hitchhiker's Guide to the Galaxy around age ten, and about Neal Stephenson's The Diamond Age influencing Sorcero's technological imagination. Tolkien supplied questions about power and its responsible use. Douglas Adams supplied the danger of answering before one has framed the right question. Das even writes her journal in Elvish, a fact that makes most founders' morning-page rituals look under-designed.
The literary appetite does not make her operating philosophy dreamy. In an interview, she praised the combination of pressure, speed and carefully designed experiments, and said she had her team read Frank Slootman's Amp It Up. Her version of moving quickly still requires intention. An experiment is only useful if its design can tell you something. That insistence echoes the 300 interviews: curiosity must eventually become a decision.
The instinct may be familial. Das has described herself as the daughter and granddaughter of women scientists. Her mother is a solid-state physicist; her grandmother was a geophysicist and mathematician. As a child, Das played in her mother's laboratory and meddled with experiments run by a computer important enough to occupy its own room. The enduring lesson she draws from the women in her family is concise: persist.
“The only way to win is to persist.”Dipanwita Das
Persistence, in her telling, is not decorative grit. It is remaining at the table when skepticism has decided in advance who looks capable of scale. Das has spoken publicly about the gendered doubts she encountered while fundraising and the value of indexing toward investors and colleagues who support women leaders. Her answer has been continued work, a widening customer base and the steady replacement of claims with operating evidence.
Capital catches up with the thesis
By 2021, Sorcero had announced a $10 million Series A. Das was also listed as an inventor on a patent for cross-context natural-language model generation, one piece of a patent portfolio her public biographies connect to biomedical AI and large language models. The company kept narrowing its language from general knowledge management toward medical intelligence, scientific communications and evidence workflows.
In November 2025, Sorcero announced a $42.5 million Series B led by NewSpring Growth, bringing its announced total capital raised to $59 million. The company said the money would expand its work across medical affairs, safety, scientific communications and medical devices. Das framed the ambition as a unified intelligence platform for the precision-medicine era. The important word is unified: the original problem of scattered knowledge remains visible even as the vocabulary and models have changed.
The round followed Sorcero's first appearance on the Inc. 5000, where it ranked No. 684 in 2025. In June 2026, Das was named to the 100 Women in AI list produced by Flybridge and XFactor Ventures. Awards and financing are tidy punctuation marks, but they can make a founder's path appear straighter than it was. Sorcero's story contains a broad initial market, years of research, customer education, a vertical turn and repeated revisions to how the company explained what it sold.
The stealable part
What can another builder take from Das? Begin with the expensive human bottleneck, not the theatrical capability. Interview enough people that their work stops looking like a persona slide. Treat focus as an allocation of learning, because every new vertical creates another language, workflow and trust problem. And when a product enters a regulated field, make its evidence inspectable. A clever answer without a visible path back to the material is merely a confident stranger.
Das's career has repeatedly returned to infrastructure: a training system on which leaders can grow, a communication platform through which evidence can travel, a software layer upon which experts can ask better questions. Infrastructure rarely gets the glamorous sentence. It becomes visible only when it fails. That may explain her attraction to it. Build the dependable layer, and other people get to do more ambitious work above it.
There is also a gentler lesson in the Elvish notebook. Large systems need imaginative people, but imagination earns its keep through attention. Sorcero's central wager is that the distance between evidence and action can be shortened without erasing the expert in between. Das arrived at that wager through policy work, a dinner conversation, hundreds of interviews and years of iteration. The future may be written in code. First, however, somebody has to listen closely enough to know what it should say.