You type a query about a novel photonic sensor material into a search bar. The results that come back are not academic papers, though they are built on them. Instead, you get a ranked list of industrial manufacturers, annotated with their recent patent filings and the names of researchers whose work suggests a technological fit. The interface is clean, the typography modern, but the promise is ancient: turning knowledge into a map. This is the first moment inside Lupa, an AI platform built to find a commercial path for discoveries that often have none [hellolupa.com, retrieved 2026].
The Wedge Between Discovery and Deal
Lupa, built by Zurich-based Prismus, operates in the narrow, high-stakes gap between a scientific breakthrough and a signed contract. Its substrate is the fragmented universe of research publications, grants, patents, and corporate activity. Its agents are algorithms designed to crawl this data, drawing connections between, say, a lab's paper on quantum error correction and a company's recent funding round for quantum hardware [Dealroom, October 2026]. For its customers, deep-tech startups in fields like photonics, quantum computing, and life sciences, the product promises to replace a manual, intuition-driven business development process with a systematic, AI-driven scan of the horizon. The company reports it has secured its first 30 paying customers, all Swiss deep-tech firms, who are using it for customer outreach and market strategy [Tech.eu, October 2026] [Startup.eu, October 2026].
A Team Built for the Translation
The founding team reflects the bilingual fluency the product demands, combining scientific rigor with commercial pragmatism. CEO Elisabeth Blyumin brings a background in computational physics from ETH Zurich, while CTO Timen van Gelderen and Chief AI Officer Gonçalo Hora de Carvalho handle the engineering and model layers [Tech.eu, October 2026]. The appointment of Peter Kempin as chief revenue officer signals an early focus on commercial traction [Tech.eu, October 2026]. This structure suggests a company aware that its technical prowess must be matched by an equally sharp go-to-market motion from day one.
The company's initial financial runway comes from a CHF 150,000 (approximately $182,000) grant awarded by Swiss accelerator Venture Kick in October 2026 [Biztoc, October 2026]. This non-dilutive capital is earmarked for advancing the Lupa platform and scaling within deep-tech markets, a typical use for early-stage grant funding in the European ecosystem.
The Traction and the Tightrope
The early claim of 30 paying customers is a strong signal for a product launched in 2025, suggesting Lupa is solving a tangible, immediate pain point. These are not free-tier users; they are companies willing to pay for intelligence that could dictate their next sales hire or partnership pitch. The concentration in Switzerland's robust deep-tech scene provides a dense, initial proving ground.
Yet, the path ahead is a tightrope walk between depth and scale. The platform's value is intrinsically tied to the quality and specificity of its data connections. The risks are not about competition from a named rival, but about the fundamental challenges of the category itself.
- Data fidelity. An AI agent's recommendation is only as good as the data it ingests and the logic it applies. A missed patent or an erroneous link could send a sales team on a costly wild goose chase. The platform must achieve near-perfect accuracy to maintain trust.
- Market breadth. Success with Swiss quantum and photonics firms is one thing. Translating that to different scientific disciplines or geographic markets requires adapting the underlying models and data sources, a non-trivial expansion effort.
- The human factor. The product aims to augment, not replace, the domain expert. Its ultimate test will be whether a seasoned business development director in deep tech finds its leads not just novel, but genuinely superior to their own network and research.
What the Next Check Must Prove
The Venture Kick grant is a start, but the next institutional funding round will need to answer a harder question: can Lupa move from being a useful tool for a niche clientele to becoming an essential platform that defines how deep-tech commercialization is done? Scaling will require significant investment in data acquisition, model refinement, and a sales team capable of speaking the language of both the lab and the boardroom.
For now, Lupa operates on a premise that feels both obvious and overdue. In an age where scientific publication accelerates but commercialization lags, it asks a simple, cultural question: what if the hardest part of building a deep-tech company wasn't the science, but simply knowing who to call? The platform is a bet that the answer lies not in more networking events, but in teaching a machine to read the room, a room built from every patent, paper, and funding announcement ever filed [Lupa - Science and Industry Intelligence Platform, retrieved 2026].
Sources
- [Tech.eu, October 2026] Prismus lands €163K to bridge the gap between scientific discovery and industry | https://tech.eu/2026/10/30/prismus-lands-eur163k-to-bridge-the-gap-between-scientific-discovery-and-industry/
- [Dealroom, October 2026] Prismus raises $182K grant for AI platform linking science to industry | https://app.dealroom.co/news/note/prismus-raises-182k-grant-for-ai-platform-linking-science-to-industry
- [Startup.eu, October 2026] Swiss deep-tech startup Lupa (Prismus) secures CHF 150,000 from Venture Kick for AI research platform | https://www.startup.eu/investments/prismus-150-k-10-2026
- [hellolupa.com, retrieved 2026] Lupa - Science and Industry Intelligence Platform | https://hellolupa.com/
- [Biztoc, October 2026] Prismus lands €163K to bridge the gap between scientific discovery and industry | https://biztoc.com/x/8b244f6da1fbdfc8
- [Lupa - Science and Industry Intelligence Platform, retrieved 2026] Lupa researches and works across publications, grants, funding news, patents, companies, labs, citation graphs and co-author networks | https://hellolupa.com/