In a factory where every product is a little different, the most expensive sensor is often the human eye. It’s a tired, expensive, and error-prone sensor, especially when the assembly instructions change with every batch. Sentinus AG, a 2023 spin-off from ETH Zurich, is building a camera system to replace that final manual glance, betting that a visual AI can be trained to spot mistakes in the messy, low-volume world of bespoke manufacturing [Sentinus.ch, 2025].
The wedge into high-mix manufacturing
Sentinus targets a specific and stubborn corner of industry: high-mix, low-volume (HMLV) manual assembly. This is where products are built in small batches or even as one-offs, like specialized medical devices, aerospace components, or complex industrial machinery. Traditional, fully automated vision systems struggle here because they require consistent, predictable parts and lighting. The cost of programming and maintaining them for constant change is prohibitive. Sentinus proposes a simpler add-on: a smart camera that watches the worker’s hands, compares the assembly to a digital instruction set, and flags discrepancies in real time [Sentinus.ch, 2025]. The product is pitched as a layer on top of existing digital work instruction systems, aiming to close the loop between guidance and verification.
A team built for industrial AI
The company’s technical roots are its clearest asset. Co-founders Jonas Conrad and Felix Schnarrenberger both hail from ETH Zurich, with backgrounds in mechanical engineering, process systems, and robotics [Crunchbase, 2026]. Conrad remains a research associate at inspire AG, an ETH-affiliated transfer center for production technologies, suggesting a pipeline from academic R&D to industrial application [inspire.ch, 2026]. Schnarrenberger brought practical machine learning experience from a prior role at cable and connectivity specialist HUBER+SUHNER [Crunchbase, 2026]. It’s a profile common to Swiss deep tech: strong on engineering rigor, lighter on commercial track record. Their early backing came from Venture Kick, a Swiss non-profit that provides seed funding and coaching to university spin-offs, which invested $173,000 in late 2024 [Seedtable, 2026].
| Role | Name | Background |
|---|---|---|
| Co-Founder | Jonas Conrad | Research Associate, ETH Zurich / inspire AG [Crunchbase, 2026] |
| Co-Founder & Technical Lead | Felix Schnarrenberger | MSc Robotics, ETH Zurich; ex-ML Engineer, HUBER+SUHNER [Sentinus.ch, 2026] |
| Operational Lead | Constantin Herbst | Not specified in public sources |
| Computer Vision Engineer | David Filiberti | Not specified in public sources |
The unit economics of a caught error
The bet rests on a simple calculation. In quality-critical assembly, a defect that slips through can trigger a cascade of costs: warranty claims, recalls, scrap, and reputational damage. Catching it at the station is cheapest. Sentinus’s cameras are meant to be the final, tireless inspector. The company is running a pilot program to prove the case, though it has not yet publicly named any customers or published results [Sentinus.ch, 2025]. The path to scale will depend on demonstrating a clear return on investment that beats the status quo of human inspectors and post-assembly testing.
Back of the envelope: If a single critical assembly error costs a manufacturer €10,000 in rework, scrap, and delay, and a Sentinus system prevents just one such error per month, it pays for a typical hardware-plus-software deployment in a matter of quarters. The real value, however, is in preventing the error that never happens because the worker gets real-time guidance.
Where the vision could blur
For all its technical promise, Sentinus faces a crowded field of well-funded incumbents and emerging startups. The competitive landscape isn't about a lack of vision systems; it's about which one a production manager will trust and buy.
- The industrial giants. Companies like Cognex and Keyence dominate the market for robust, high-speed machine vision. Their systems are the gold standard for high-volume production lines. Sentinus must convince customers that its AI-driven, flexible approach is better suited for volatile HMLV environments than trying to adapt a Cognex system.
- The platform players. Larger manufacturing execution system (MES) and IoT platforms are increasingly bundling AI-powered quality modules. A buyer might prefer an integrated solution from their existing software vendor over a standalone hardware add-on.
- The proof gap. The most immediate hurdle is moving from pilot to paid deployment. Without public case studies or named reference customers, the value proposition remains theoretical. The Venture Kick grant validates the technology at a prototype level, but the next round will need to be justified by commercial traction.
The company to beat is Cognex. Its systems are on millions of production lines worldwide, synonymous with reliability. For Sentinus to win, it must prove that in the chaotic, changeable world of high-mix manufacturing, flexibility and ease of retraining are more valuable than raw speed and precision.
Sources
- [Sentinus.ch, 2025] Sentinus homepage and product description | https://sentinus.ch/
- [Crunchbase, 2026] Jonas Conrad profile | https://www.crunchbase.com/person/jonas-conrad
- [inspire.ch, 2026] Jonas Conrad team profile at inspire AG | https://www.inspire.ch/en/about-us/team/conrad-julian/
- [Sentinus.ch, 2026] Team page listing Felix Schnarrenberger | https://www.sentinus.ch/team.html
- [Crunchbase, 2026] Felix Schnarrenberger profile | https://www.crunchbase.com/person/felix-schnarrenberger
- [Seedtable, 2026] Sentinus funding information | https://www.seedtable.com/startups/sentinus