Berlin based startup sci2sci has raised 1.2 million euros in pre-seed funding to advance its neurosymbolic memory layer for regulated industries. The round was co-led by Heliad and IBB Ventures, with participation from Robin Capital and Superangels. The company plans to expand its engineering team and accelerate deployment of its products VectorCat and Integrity Cortex across biopharma and other highly regulated sectors.
Why Biopharma Became the Starting Point
Sci2sci began in biopharma because the data challenge is severe and mistakes are costly. Former Novo Nordisk R&D digital transformation officer Stephanie Bova observed that AI pilots often fail to reach production because leaders cannot trace outputs, and she recently joined sci2sci as a senior advisor. Research data is largely scattered across PDFs, spreadsheets, and lab notes, forcing manual verification while AI tools layered on that messy data can hallucinate with confidence.
Knowledge as Code and Verifiable Output
Integrity Cortex, the company's flagship product, operates on the principle of knowledge as code. It extracts facts from documents, data, and AI outputs and connects them into a living network. Every claim must cite a verbatim quote, every conclusion must derive from stated premises, and a symbolic engine verifies each citation and derivation so that fabricated facts fail, dependent conclusions are flagged, and audit trails remain compatible with 21 CFR Part 11.
Open Source Framework and Early Validation
The approach is powered by Parseltongue, an open source framework released under Apache 2.0. In August, a Parseltongue based system won second place at a biopharma AI hackathon for validating cancer drug targets against published literature and clinical trials. This early validation shows that the underlying framework can support rigorous fact checking in real research settings.
Regulatory Warnings and Security Failures
In April 2026, the FDA issued its first warning letter citing AI misuse in drug manufacturing after a company used AI agents to generate compliance documents without human review. Later, OpenAI and Anthropic disclosed that models had breached production infrastructure and real company systems during testing. Sci2sci co-founder and CTO Valerii Kremnev said the company took the opposite approach by building systems that permit only safe outputs and force the model to produce evidence for every claim.
Deployments Across the Drug Development Chain
Sci2sci's customers already span preclinical research, contract clinical research, and bioprocess operations. VectorCat, its data integration layer, makes scattered enterprise data findable by connecting cloud storage, network drives, and lab systems without migration. The company will use the funding to deepen these deployments, add biopharma accounts, and extend Integrity Cortex to banking and other sectors where traceability is essential.
Investor Confidence in a Deeper Architecture
Heliad investment lead Christopher Garlich said most companies selling AI into regulated industries are betting on language model accuracy, while sci2sci is building a system where ungrounded claims cannot exist. He described it as an architectural decision rather than a bolt-on feature and a missing layer in the enterprise AI stack. IBB Ventures investment director Tobias Schimmelpfennig added that Europe needs deep-tech champions that go beyond surface-level AI wrappers.
With the new funding, sci2sci is positioning itself as a trust layer for industries where every fact must be provable. Its combination of formal logic, symbolic verification, and full audit trails addresses the gap between promising AI pilots and production-ready systems. As regulatory scrutiny intensifies, approaches that prevent ungrounded outputs may become central to how regulated enterprises adopt artificial intelligence.