Welcome to this week's issue of Deep Tech Brief.
Every Saturday, we break down deep tech companies building genuinely hard things, in plain language. No jargon, no hype.
This week's three companies: robot traffic control in Berlin, plastic marking in St. Gallen, laser spectroscopy in Zurich.
Let’s dive in.
1. SyncroBot: a traffic controller for robots that were never built to share a floor
Warehouse robots have an odd adoption problem. The hardware is mature and the business case is well understood, yet Interact Analysis projects that by 2030 only 13% of warehouses will have deployed even one fulfilment AMR. Something other than the robots is the blocker.
That something is integration. Every vendor ships its own fleet manager and its own control interface, so putting two brands on one floor means either custom development or a decade of vendor lock-in. The German industry standard VDA 5050 exists to fix this, and it helps, but conformance is uneven and vendors still expose their real capabilities through proprietary APIs.
SyncroBot builds middleware that sits above the fleet, coordinating tasks, routing and traffic rules across mixed brands, syncing a shared map to every vehicle in real time, and letting operators take remote control of any robot through live camera feeds. One API instead of one per manufacturer.
The clever part is what SyncroBot leaves out. Incumbents like KUKA's fleet software, KINEXON and SYNAOS also orchestrate mixed fleets, but they lean on integration with warehouse management, ERP and MES systems, which is what takes months. SyncroBot skips it: a task fires from the app, a button, or a barcode scan. The bet is that the value is in coordination, and the plumbing can wait.
SyncroBot is a spin-off of the Technical University of Berlin, founded by Dr. Julian Maas (CEO), Naseem Ashley Kasraee (COO) and Linus Pfoch. Kasraee notes that robotics still reads as scary to operators, so the team runs a showroom at Fraunhofer IFF in Magdeburg where prospects drive real robots first. By Maas's account the company is financed and focused on commercialisation, with headline figures of five-day deployment, 80 to 90% lower deployment costs and up to 40% better fleet utilisation.
Worth watching, because "skip the integration" is a genuinely different answer to the industry's slowest problem. The risk: orchestration middleware is crowded with well-funded incumbents, and as VDA 5050 matures the translation layer gets easier for everyone to build. Defensibility here is commercial speed, not code.
Learn more: syncrobot.io
2. matriq: giving every plastic part a name at the moment it is formed
Traceability rules are arriving faster than the plastics industry can mark parts. The EU's Ecodesign for Sustainable Products Regulation took effect in July 2024 and brings the Digital Product Passport with it, starting with batteries in February 2027 and phasing across most physical goods by 2030. Every passport needs a unique identifier physically attached to the product. Counterfeiting adds to the pressure: the OECD and EUIPO put global trade in fake goods at roughly USD 467 billion in 2021.
Today that identifier gets added afterwards, at a laser station downstream of the molding machine: an extra step, a handling robot, floor space, validation, and sometimes dust a cleanroom cannot tolerate.
matriq's DynamicMold technology moves the marking inside the mold. An 8 mm insert sits in each cavity, installed much the way a temperature sensor is, wired to a controller on the mold. As the part forms, the insert writes a 2D DataMatrix code into the plastic: a serial number, or a time stamp accurate to the second. The identity is created in the same instant as the part itself.
The edge is that this adds nothing: no extra cycle time, no footprint, no consumables, no particles. It also produces a mark that is harder to fake, because the code comes from the tool itself and is tied to a specific mold, cavity and shot rather than applied later by equipment anyone can buy.
matriq AG was founded in St. Gallen in October 2019 as a spin-off of OST, the Eastern Switzerland University of Applied Sciences, by André Bernard, Cornelia Nef, Mathias Mächler and Klaus Dietrich, who is CTO. The technology is patented, it won the Swiss Technology Award in 2023, two products have shipped, and backers include Startfeld, CADFEM and kickfund. Dietrich says the company is pre-Series A.
One to watch, carefully. The technology is proven and the regulatory tailwind is real, but adoption is gated by something matriq does not control: molds are expensive, replaced rarely, and the insert must be designed in. Dietrich's own answer about his hardest challenge, that it took far longer than anyone anticipated, is the honest signal.
Learn more: matriq.ch
3. Enantios: measuring the shape of molecules AI can now design faster than anyone can test them
More than half of drugs in use are chiral: they exist in two mirror-image forms built from identical atoms, and the two behave differently in the body. That is why thalidomide happened, and why the FDA issued its stereoisomer policy in 1992. Regulators now expect a drug's handedness to be characterised.
The trouble is that design has accelerated and measurement has not. By early 2026, industry trackers counted more than 170 AI-originated drug programmes in clinical development, up from roughly two dozen in late 2023. The conventional tools each carry a hard prerequisite: chiral HPLC needs a purified reference standard developed molecule by molecule, and X-ray crystallography needs a crystal, which peptides, oligonucleotides, antibody-drug conjugates and most large biologics refuse to provide. As CEO Carin Lightner puts it, candidates die on the computer screen once someone tries to make them.
Enantios uses Raman optical activity: shine circularly polarised laser light at a molecule in solution and measure the tiny difference in how it scatters right-handed versus left-handed light. That difference reports the molecule's three-dimensional structure directly. No crystal, no chiral HPLC, no reference standard.
The hard part is that the effect is minuscule, roughly a thousandth of the ordinary Raman signal, which is why ROA sat in specialist academic labs for decades. The reference instrument, BioTools' ChiralRAMAN, was the first and for years the only dedicated ROA spectrometer sold, and it takes two to five hours for a good protein spectrum. Building on Lightner's doctoral work at ETH Zürich, Enantios is attacking usability: small enough for a glove box, three excitation wavelengths for awkward samples, and a service model so customers can send molecules in rather than buy hardware.
Enantios AG is an ETH Zürich spin-off incorporated in December 2022 by Lightner and Dr. Roman Wyss (CTO). It has raised over CHF 1.8 million from S2S Ventures and Zürcher Kantonalbank and ranked 16th in the Top 100 Swiss Startups of 2025. By Lightner's account the Gen 1 instrument is running at customer sites with top pharmaceutical clients, and roughly $5 million is now being raised.
Worth watching, because chiral analysis is a real and worsening bottleneck and Enantios is selling into it with a shipping product rather than a promise. The risk is the classic one for single-technique instrument makers: ROA either becomes a standard line in pharma QC protocols, or stays a specialist add-on. No customers are named publicly, so traction is a founder claim.
Learn more: enantios.com
Know a deep tech company we should break down next? Hit reply and tell us. The best editions come from real tips.
See you next Saturday.
Brought to you by Alluvium Media, the #1 deep tech content studio, helping founders turn complex science into video that raises money and wins customers.