sci2sci raises $1.4M pre-seed to build trusted AI for regulated industries
What's the deal? Berlin-based startup sci2sciDealroom has a profile for this one. Try Dealroom → has raised €1.2 million (roughly $1.4 million) in pre-seed funding to expand software that organises and verifies data in regulated industries. The round was co-led by HeliadDealroom has a profile for this one. Try Dealroom → and IBB VenturesDealroom has a profile for this one. Try Dealroom →, with participation from Robin CapitalDealroom has a profile for this one. Try Dealroom → and SuperangelsDealroom has a profile for this one. Try Dealroom →.
What's the endgame? Founded by Angelina Lesnikova and Valerii Kremnev, sci2sci helps companies connect fragmented information and verify the evidence behind data and AI-generated outputs. It started in biopharma, where research and regulatory records are often scattered across PDFs, spreadsheets, lab notes, and disconnected systems.
The company's Integrity Cortex links claims to their sources and verifies citations, producing an auditable record designed to meet 21 CFR Part 11 rules for electronic records. It is built on Parseltongue, a framework sci2sci open-sourced this year under the Apache 2.0 licence. A second product, VectorCat, connects data across cloud storage, network drives, and lab systems without forcing companies to migrate it.
Why now? Fragmented data can make it hard to trace findings back to their sources during audits, and AI tools can compound the problem when working with incomplete information.
What's next? Sci2sci already works with customers in preclinical research, contract clinical research, and bioprocess operations. It plans to use the funding to expand engineering, deepen deployments, reach more biopharma customers, and extend Integrity Cortex into other regulated sectors, including banking.
Kremnev framed the approach as building trust into AI from the start. "Instead of shipping capabilities first and patching trust and security afterward, we built Parseltongue and Integrity Cortex to permit only safe outputs," he said, adding that the systems force a model "to produce evidence for every claim it makes."
The signal: The round sits above the midpoint for deals of its kind, in the 28th percentile by size. As enterprises push AI into audit-heavy fields, startups selling verification and traceability — rather than raw capability — are carving out a distinct niche.
Read more: Tech.eu
Image credit: Generated with Gemini
