- To secure €1.2 million in pre-seed funding, the company aims to enable tracing of AI-generated claims back to their original data.
- The funding round was completed shortly after OpenAI and Anthropic admitted that their AI systems had accessed real companies without permission.
- Heliad and IBB Ventures led the funding round, while Robin Capital and Superangels also participated.
Angelina Lesnikova obtained her PhD by investigating the processes of memory formation and decay, beginning in a neuroscience laboratory in Helsinki and later examining how the SARS-CoV-2 virus enters brain cells. She now applies her traceability expertise to enhance AI-generated insights for biopharmaceutical companies.
The Berlin-based startup she runs, sci2sci, has recently secured €1.2 million in pre-seed funding to advance this work.
The round was co-led by Heliad, a firm whose portfolio includes the cybersecurity startup Aikido and AI21 Labs, as well as IBB Ventures, which has invested in approximately 310 Berlin companies, including ARC Intelligence, Bounti, and Vara. Robin Capital and Superangels also took part.
When a black box turns into a risk
The fact that the funding round took place at that time is significant. In July, OpenAI announced that one of its models had escaped from a testing environment and gained access to Hugging Face’s production systems.
A few days later, Anthropic stated that three of its Claude models had done the same thing, reaching out to three actual companies during security tests.
Two of those companies only became aware of it when Anthropic informed them. In April, the FDA issued its first warning letter regarding the misuse of AI in drug manufacturing after a company admitted that it had skipped a necessary process check because its AI agent had not flagged the step. That company has since stopped making drugs.
“When the two labs that market themselves as the most safety-conscious both have models breaking into real companies in the same month, it’s a process design flaw. We built Parseltongue and Integrity Cortex to permit only safe outputs. A model’s lockpicking skills are useless when there’s no door,” said Valerii Kremnev, sci2sci’s co-founder and CTO, who spent years building trading infrastructure at Devexperts, Yandex, and Delivery Hero.
From brain cells to knowledge graphs
The main product of sci2sci, Integrity Cortex, does not use language models to summarise company documents. Rather, it extracts each fact and stores it in a formal logic system.
Each claim must be supported by a direct quote from the source; every conclusion has to be based on clear premises, and a symbolic engine verifies each step. If a model invents a fact, the system will detect it and flag any conclusions related to it.
The company’s second product, VectorCat, functions as a data catalog by linking lab notebooks, cloud storage, and network drives without moving the data. This arrangement uses Parseltongue, a framework that sci2sci released this year under the Apache 2.0 licence as open source.
A team that used Parseltongue came in second in a biopharma AI hackathon by comparing cancer drug targets against clinical trial literature.
The market is very competitive
The London-based company Pharosyn has raised $3 million to trace pharmaceutical intelligence back to its sources, while the Swiss startup Rivia has secured €3 million to gather biotech clinical trial data. French startup White Circle, founded by former employees of OpenAI, Anthropic, and DeepMind, has raised $11 million to detect AI mistakes after deployment.
Sci2sci takes a different route. Instead of examining the results afterwards, it aims to prevent unsupported claims from occurring in the first place and provides an audit trail that complies with 21 CFR Part 11 standards for FDA submissions.
“Most companies selling AI into regulated industries are betting that language models will be accurate enough. Sci2sci is making a different bet: build a system where an ungrounded claim simply cannot exist,” said Christopher Garlich, Investment Lead at Heliad.
Tobias Schimmelpfennig, investment director at IBB Ventures, added: “Angelina and Valerii are combining genuine life sciences domain depth, cutting-edge AI research, and enterprise-scale systems engineering.”
The company has recently appointed Stephanie Bova, who previously served as corporate vice president and digital transformation officer for Novo Nordisk’s R&D department, as a senior advisor.
Sci2sci’s customers are involved in pre-clinical research, contract clinical research, and bioprocess operations, but the company has not disclosed its headcount or revenue.
What comes next
The new funding will be used to expand the engineering team and will enable Integrity Cortex to enter the banking sector, as document tracking is also relevant there.
The global AI governance market will grow from $890 million in 2024 to $5.78 billion by 2029. The EU AI Act, which imposes fines of up to €35 million on non-compliant high-risk systems, provides regulated industries with a clear deadline for addressing these issues.
Since banking has its own complex, country-specific compliance requirements, the system developed for submission to the FDA will need to demonstrate its ability to adapt before sci2sci’s next funding round.
