What if every experiment only had to happen once?

The life sciences produce a lot of data, but much of it is spread out in documents, databases, and internal systems. Years later, researchers might remember an experiment but struggle to find its results, link them to other work, or know which information is reliable.

Sci2Sci Co-founder and CEO, Angelina Lesnikova saw this problem herself as a neuroscientist. She noticed that labs often focus on single proteins, pathways, or processes, but biology is all about connections.

This inspired her to create Sci2Sci. The platform brings together scattered data of all kinds and turns it into organized, reliable knowledge that both people and AI can use.

Angelina’s goal goes beyond helping scientists find information quickly. She wants to ensure valuable research is never lost and help life sciences move faster from experiments to breakthroughs.

About Sci2Sci

Sci2Sci was founded by Angelina Lesnikova (CEO) and Valerii Kremnev (CTO) who left their respective fields to tackle what they see as one of life sciences’ biggest bottlenecks: knowledge management. The team believes that the way knowledge is currently stored, connected and used is holding back progress across healthcare, longevity, quality of life and sustainability. And solving that problem requires more than research or product development alone.

Sci2Sci combines the two, conducting research into how life sciences can work better while turning those ideas into products tested against real-world constraints. The goal is to build practical infrastructure that helps scientific knowledge move faster, and ultimately accelerates progress across the industry.

From politics to neuroscience to entrepreneurship

Angelina didn’t follow a typical path into the life sciences. She started her first degree in international politics at 16.

After realizing that politics was not the field she wanted to spend her career in, she discovered that what interested her was intelligence in itself, the human inner world, imagination, creativity and our capacity for generating completely new ideas.

She became especially fascinated by the possibility of brain-computer interfaces advanced enough to pick up human thought and communicate it directly.

There was one problem.

In order to create something like this, she had to understand how the brain worked. So, she turned her attention to neuroscience.

When she was halfway through her PhD, Angelina reached a frustrating conclusion: even neuroscientists don’t fully understand how the brain works. However, by then she had become so deeply involved in the field and grown more interested in another basic problem she saw recurring throughout scientific research.

Although science was generating vast quantities of knowledge, researchers were only able to concentrate on fairly narrow aspects of it. It was even difficult to keep up with the research that had been published in a single specialized field. And no individual researcher could really understand how discoveries across the entire biological system were connected.

AI has changed what’s possible, but it hasn’t solved the core data problem.

It would not be enough to provide a model with all the scientific papers, internal experiments, and datasets ever produced if reliable insights were to be expected. For that, the knowledge first has to be organized, linked, and made comprehensible to machines. This realization became one of the foundations of Sci2Sci.

“Nothing in the brain or in biology works in isolation.”

Listening before building

Angelina and the team consulted over 100 professionals in the biotechnology and pharmaceutical industries before deciding on exactly what Sci2Sci should become.

The discussions helped the team identify where companies struggled and how much skepticism they could expect when trying to sell technology to that industry.

Angelina points out that many companies have put a great deal of money into data projects only to find that the results were not as expected, which is why there is a reasonable level of distrust towards both new and established vendors.

Sci2Sci decided to look more closely at the problem.

Instead of offering companies a predetermined solution, the team used those initial conversations to learn how different kinds of data actually travel through research organizations, where projects get stuck, and what compliance and data teams need before they can take on new technology.

That approach is especially important when targeting an industry that is not only highly regulated but also traditionally difficult for very early-stage companies to enter.

Angelina explains that people often told her and co-founder Valerii that it was "crazy" to begin in the life sciences field, where large organizations are often unwilling to even talk to startups, let alone provide them with access to essential data infrastructure.

It was exactly this complexity that made this problem worthwhile.

Sci2Sci now provides services to customers worldwide, including businesses in the United States and other parts of Europe. Even though the company is based in Berlin, Angelina states that its customer base has always been international, a situation made possible by the fact that the first discovery discussions were carried out digitally rather than restricting the company to its local environment.

“We really made an effort to understand what our customers wanted.”

Turning scattered data into usable knowledge

Angelina describes the data problem in life sciences with a simple analogy: imagine a stadium filled with books, all thrown into one enormous pile. The information is there, but finding the right piece, or understanding how it connects to everything else, is incredibly difficult. That is what Sci2Sci wants to fix.

Research data can be scattered across electronic lab notebooks, cloud storage, presentations, and other systems, often with no clear links between them. VectorCat connects to that data where it already lives, checks it for potential compliance issues, and then indexes and enriches it so both people and AI can understand what exists and how it relates.

The result is a way to organize the stadium. Sci2Sci can reconnect parts of research projects that have become separated across different systems, helping teams find and reuse work that might otherwise be lost.

“In any single lab I worked in, worked with or interviewed, it was way easier to redo an experiment than to find data from the previous year.”

But Angelina’s ambition goes further. Sci2Sci’s Integrity Cortex, named after the cortex in the human brain, aims to move beyond documents altogether, extracting individual facts while keeping them connected to their original sources. Those facts can then be linked, checked for inconsistencies and updated as new knowledge emerges.

Instead of searching through piles of documents, researchers could work with a connected network of verifiable knowledge.

“We are moving from one document to one fact, one verifiable fact, as the unit of knowledge.”

Building a memory for AI inspired by the human brain

Angelina's approach to AI goes back to 2018, the year she met co-founder Valerii.

At the time, Angelina was giving popular science lectures and carrying out her academic work. After one of her lectures on memory and the human brain, which Valerii attended, they started talking.

As a software engineer, Valerii attended the lecture because he was interested in whether principles from neuroscience could improve artificial intelligence.

What impressed Angelina most was the depth of Valerii's questions. Although scientists tend to specialize, Valerii aimed to understand the brain as a whole system. It was many years before that way of thinking affected the architecture of Sci2Sci's Integrity Cortex.

As Angelina points out, most current approaches to AI memory still resemble databases: information is stored as separate records or fragments and, when the model needs to answer a question, the system retrieves the pieces it considers most relevant. But those pieces can be disconnected and contradictory.

Human memory works differently.

It is a network that constantly changes as new knowledge comes in, since new information is not just added to a fixed archive but instead alters the relationships between things that are already known. Sci2Sci’s Integrity Cortex is based on the same principle.

The system takes facts from documents and data assets, tracks how they relate to their sources, and establishes connections between them; when new information comes in, it compares it with existing knowledge and updates the parts of the network affected by the change.

It turns a process that currently involves a great deal of manual work into one that is much more dynamic.

For example, if a pharmaceutical company alters the dosage in the clinical trial protocol, that one change would have to be incorporated into the statistical analysis plan, the materials provided to the investigators, and a large number of other documents. At present, it can take weeks to identify and reconcile all the references.

Since Integrity Cortex knows every instance where the underlying fact is recorded, it can instantly tell what information needs updating.

“In the human brain, memory works nothing like a database.”

In life sciences, AI has to be verifiable

Sci2Sci's approach clearly differs from the general trend of introducing unguarded generative AI into every industry. The life sciences should not have a "move fast and break things" attitude, says Angelina.

In a low-stakes situation, a hallucinated answer might be inconvenient; however, in the life sciences, incorrect information entering a clinical procedure can have much more serious consequences.

Frontier AI models, by themselves, therefore do not address the problem Sci2Sci addresses. Angelina maintains that the missing element is corroboration, that is, deciding which information is correct when an organization has conflicting versions, experimental results, or documents.

The Sci2Sci system can obtain information from multiple sources and then check those facts against each other. If the information does not agree, the inconsistency appears rather than being quietly included in future outputs.

The more AI organizations deploy, the more important this becomes.

Errors can accumulate if the underlying knowledge the models are based on is unreliable. The more processes that rely on that information, the farther a wrong assumption can spread.

With Sci2Sci, Angelina and Valerii are building an infrastructure in which AI outputs are linked to evidence and contradictions can be detected rather than suppressed.

The aim is not merely to increase AI’s capabilities. It is to make it trustworthy enough for an industry where accuracy matters.

“‘Move fast and break things’ means they actually break things. That’s an approach you can afford elsewhere, but not in life sciences.”

Knowing what we don't know

One aspect of the problem is making existing scientific knowledge searchable.

It might be even more effective to understand what is lacking.

When specific facts become apparent and are linked within an organization, researchers can spot the gaps between them. Rather than just investigating what the company currently knows, they can consider what it still needs to find out and determine which experiment or data point would advance that understanding.

For Angelina this has consequences for the basic process of scientific discovery.

Research teams tend to have a narrow understanding of the knowledge at their disposal. Relevant evidence might be located elsewhere within the organization, even if they are unaware of it. Other results may challenge their assumptions. In other cases, key information has not yet been gathered.

A connected knowledge layer can make those distinctions clearer.

It would then be possible for researchers to form hypotheses based on a much larger amount of evidence, test them with greater confidence, and make more intelligent decisions about which experiments to carry out next.

The aim isn’t to take scientists’ judgment away, but to give them a more complete picture so they can do what they want to do.

“It is essential to realize not only what we know but also what we do not know.”

Accelerating the life sciences

Angelina has set an ambitious target for Sci2Sci over the next five years, stating that she wants its technology to be used by at least 80 percent of the world's top 100 life sciences companies.

It’s not just about gaining market share. It’s about speeding things up.

As Angelina points out, getting a new drug onto the market can still take 10 to 15 years; this is far too long, especially since inefficient knowledge management causes researchers to waste valuable time searching for information, checking and reconciling documents, or repeating work that has already been done.

Sci2Sci argues that better knowledge infrastructure can help address this.

If researchers can locate existing experiments rather than repeat studies, understand relationships between projects, spot conflicting evidence, and identify remaining gaps, the research process could become much more efficient.

Angelina's motivations are also very personal.

She talks openly about the need for biotechnology to make rapid progress to extend both human health and lifespans within her own lifetime, which is one reason she and Valerii have decided to work in an industry that is notoriously difficult rather than one that is easier to sell.

“The end goal for us is really about accelerating the industry.”

Quick-fire round

A tool you can’t live without?

“My note taker on my phone. I’ve been recording ideas on my phone for more than a decade and now have thousands of notes saved there. Sometimes I just have random ideas, and I really want to make sure they don’t get lost.”

How do you recharge?

“When I have time, I’m working on a book. I’m actually working on a science fiction novel, which gives me the chance to step away from Sci2Sci and use my brain in a completely different way.”

Founder or start-up that you admire:

“I have a love/hate relationship with Anthropic. They build great models and terrible software. A frontier-AI, near-trillion dollar company can’t build a basic 'change email' button into their user settings - can you believe it?!”

Advice that stuck:

“This advice was not given to me personally, but is rather a phrase from a book that I keep coming back to: “If you don’t know what you want, you end up with a lot you don’t.”

Making scientific knowledge work harder

The life sciences need to make better use of the vast amount of knowledge they already produce.

Sci2Sci is building the infrastructure to make that possible: governing information before AI accesses it, connecting research scattered across different systems, and turning documents into networks of verifiable facts.

For Angelina, the opportunity goes far beyond saving scientists time. If better knowledge infrastructure can help researchers avoid repeating experiments, identify gaps sooner, and make discoveries with greater confidence, it could ultimately accelerate the entire path from research to new treatments.

And life sciences is only the beginning. Sci2Sci plans to expand into other regulated industries, with the ambition of becoming the go-to platform for enterprise intelligence. Ultimately, the team envisions a kind of “knowledge market”, where information can move between AI agents, people, departments and organizations with the speed and efficiency of financial markets — always accessible, always up to date.

As workplaces fill with both human and autonomous AI participants, coordinating intelligence at scale is becoming a new challenge. Sci2Sci wants to build the infrastructure that makes that coordination possible.

After years spent trying to understand how intelligence and memory work in the human brain, Angelina is now applying those same ideas to another challenge on a different scale: helping organizations remember what they know, connect it, and put it to work. Check out their work here, and stay tuned to see where Angelina and Sci2Sci go next.