Sci2sci Raises €1.2M Pre-Seed to Prevent AI Hallucinations in Biopharma

Berlin-based sci2sci has raised €1.2 million in pre-seed funding to build AI infrastructure designed to prevent unsupported claims and hallucinations in highly regulated industries, starting with biopharmaceutical research.

The round was co-led by Heliad and IBB Ventures, with participation from Robin Capital and Superangels. The company was founded by neuroscientist Angelina Lesnikova and CTO Valerii Kremnev.

Lesnikova's research background focused on memory formation and decay, including work on how SARS-CoV-2 enters brain cells. At sci2sci, that interest in traceability has evolved into a technology platform designed to make AI-generated insights verifiable back to their underlying data.

Turning AI outputs into traceable claims

Sci2sci's main product, Integrity Cortex, takes a different approach from conventional AI systems that simply summarise documents.

Instead, the platform extracts individual facts and stores them within a formal logic system. Each claim must be supported by a direct quotation from its source, while conclusions need to be based on clearly defined premises.

A symbolic reasoning engine then verifies the logical chain behind each conclusion. If an AI system introduces an unsupported fact, Integrity Cortex can identify the issue and flag conclusions that depend on it.

This creates an audit trail intended to make AI-generated information easier to verify and review in regulated environments.

Building a data layer for life sciences

Sci2sci's second product, VectorCat, works as a data catalogue that connects information across lab notebooks, cloud storage and network drives without requiring companies to move the underlying data.

The infrastructure is supported by Parseltongue, an open-source framework released by sci2sci under the Apache 2.0 licence.

The approach has already been tested in biopharma applications. A team using Parseltongue placed second in a biopharma AI hackathon after using the technology to compare cancer drug targets with clinical-trial literature.

Why AI traceability matters

The company's funding comes as concerns around AI reliability and autonomous systems continue to grow, particularly in industries where a misleading output can have serious consequences.

Recent incidents involving AI models accessing real-world systems during security testing, along with regulatory scrutiny around AI use in drug manufacturing, have highlighted the risks of treating AI systems as reliable black boxes.

Sci2sci is betting that regulated companies need something more fundamental than post-deployment error detection: an architecture where unsupported claims are prevented or caught before they become part of downstream decisions.

The company says Integrity Cortex also provides an audit trail designed to comply with 21 CFR Part 11 requirements relevant to FDA-regulated environments.

Expanding beyond biopharma

Sci2sci currently works with customers involved in pre-clinical research, contract clinical research and bioprocess operations.

The €1.2 million will be used to expand the engineering team and extend Integrity Cortex into banking, where complex documentation and regulatory requirements create similar demand for traceable AI-generated information.

The company is entering a competitive market alongside startups such as Pharosyn, Rivia and White Circle, which are also tackling AI reliability, pharmaceutical intelligence and enterprise AI governance.

Sci2sci's differentiator is its focus on preventing unsupported claims at the system level rather than simply checking AI outputs after they have been generated.

With the global AI governance market projected to expand significantly through 2029 and regulations such as the EU AI Act increasing pressure on companies deploying high-risk AI systems, the demand for verifiable AI infrastructure could continue to grow.