KOMPAS VC

💥 Every company in the physical world runs on a plan. Which technician drives where, which nurse works which shift, which task runs on which machine. A single driver with 15 stops already has 1.3 trillion possible routes. At 60 stops, the options outnumber the atoms in the observable universe. And most companies still try to solve that in a spreadsheet. We invested in Timefold because they close this gap. The Ghent-based team, led by Geoffrey De Smet and Maarten Vandenbroucke, has spent twenty years on exactly these problems. It started with the open-source solver OptaPlanner, now downloaded over 700,000 times a month. The numbers show up where they count, in the customers' P&L. Timefold has eliminated more than 30 million hours of overtime and removed over 10 billion kilometres of unnecessary travel for the likes of ADP, CBRE, Deutsche Bahn, Lufthansa and IBM. There's a growing assumption that AI will crack this kind of problem on its own soon, the way it has absorbed so many others. It won't. An LLM predicts plausible output, and plausible is the wrong bar for planning. A roster that sends a nurse back ten hours after her last shift looks fine on screen and is still illegal. Planning has to be deterministic, meaning the same input gives the same answer every time, with the reasoning shown. That's exactly what Timefold's solver does. That reliability matters most where the pressure is highest, and right now the pressure is labour. Skilled workers are the binding constraint across every sector we back. There are already too few technicians, nurses and drivers, and the pool is only shrinking. The one durable answer is to get more from the people you already have, and scheduling is where those gains are won. The payoff runs further than efficiency. Those 10 billion kilometres that never got driven are also fuel never burned, which puts Timefold in the rare category where productivity and decarbonisation are the same feature. Read the full breakdown of why we invested, written by Sebastian Peck ⤵️ 🔗 https://lnkd.in/eS-8srqf #kompasvc #timefold #whyweinvested

💥 Traditional satellites fail to detect continuous movement. The curve of the Earth limits ground-based radar systems. Airborne surveillance planes reach further, but they're expensive to fly and are riddled with operational complexities. That leaves a real gap for governments and companies managing busy airspace, watching contested borders, and tracking activity across large regions need continuous monitoring of what's moving in the air and at sea, anywhere on Earth. Array Labs is closing that gap with a fundamentally different approach to building radar satellites. Instead of one large, expensive satellite, Array builds smaller radar satellites made at scale that fly in a cluster and work together as a single sensor in orbit. It borrows manufacturing techniques from telecoms, semiconductors, and consumer electronics, then adds its own signal-processing software and formation-flying approach. That combination delivers more power, wider coverage, and sharper resolution than traditional space systems, at a fraction of the cost. The result is a live, continuously updated picture of Earth that captures what moves across it and above it, day or night, in any weather condition. The solution covers 2D and 3D imaging, airborne movement target indication, and awareness of other objects in orbit, all in one platform. No other commercial satellite operator offers that breadth in a single integrated system. That is why we're excited to support Array's latest raise with $21 million in strategic funding, anchored by Mitsubishi Electric (MELCO), one of Japan's largest defense and aerospace companies, alongside co-investors Catapult Ventures and Y Combinator. The round comes with a partnership. Array and MELCO will bring satellite-based aircraft tracking to defense and security customers across Asia-Pacific, pairing Array's technology with MELCO's manufacturing infrastructure and government relationships in the region. Array has already validated its formation-flying technology and won R&D contracts with Defense Advanced Research Projects Agency (DARPA), the US Navy, and the United States Air Force. "Defense and intelligence customers have been asking for this capability for years, and this round lets us move faster to deliver it," said Isaac R., COO and cofounder of Array Labs. Array is exactly the kind of company our thesis is built around; technology that makes critical systems more resilient, and does it at a scale and cost that actually reaches the field. 👏 Congratulations to Andrew Peterson, Isaac R., and the entire Array Labs team. 🔗 Full story → https://lnkd.in/g3B2enyN cc: Summit Rosenberg Sebastian Peck Andreas Winter-Extra #kompasvc #arraylabs #venturecaptial

💥 As early stage investors, we back founders building under extreme uncertainty. When a founder is intellectually curious, technically deeply competent, courageous and driven by a strong vision, the search for product market fit can generate extraordinary results. When we first met Dr David S. K., he was using synthetic biology to tackle industrial decarbonisation. His breakthrough work developing new strains of cyanobacteria, using CRISPR to supercharge their carbon fixation rate, opened up new possibilities to biomanufacture materials. As his team's ability to synthesise valuable materials grew, David turned to how to take his technology to market. Their breakthrough coincided with one of the biggest pharma booms in decades: GLP-1s. One in eight US adults now takes a GLP-1, and the first oral versions reached the market this year. But  the dose required is significantly higher compared to injections, which also impacts their cost profile.  Because oral GLP-1 is poorly absorbed and struggles to survive in the gut, a pill needs close to 50-100 times more active compound than the injection. At up to $100,000 a kilogram to synthesise, that dose size breaks the unit economics of every oral version so far. Enter General Biotechnologies. The company uses CRISPR to express therapeutic peptides directly inside a volcanic microbe discovered by 2FP &  George Church's lab at Harvard. Its thick cell wall shields the peptide from stomach acid and releases it further down the tract, so the GLP-1 can be swallowed as an edible powder at room temperature, with no injection or specialty pharmacy. The same cell wall solves the cost problem. Keeping the peptide locked inside an edible cell removes downstream purification, which can be up to 80% of manufacturing cost. As a result, General Bio can target $3 to $5 per patient per day. Three peptide products are in development for a 2027 launch, with a new facility in Fremont, California, built with Persist AI and Valentia Analytical, and North American telemedicine & distribution partners already lining up. This is the extraordinary work of the team led by Dr Uma Shankar Sagaram and Dr David Kim, a former British Army doctor and biotechnologist, with advisers from, Novo Nordisk's Moonshot factory, and George Church's laboratory at Harvard University, who count more than 25 FDA drug approvals between them. We are proud to keep backing David and his team in their $7.4M round, alongside Cartography Capital and Ritual Capital. 👏 Congratulations to Dr. David S. K., Uma Shankar Sagaram, and the whole General Biotechnologies team on this next chapter. 🔗 Read more on Axios: https://lnkd.in/gWDzBRFj Sebastian Peck Nick Gosen #kompasvc #generalbio #GLP1

Last night we brought our Full Circle dinner series to Stuttgart, sitting down with 15 founders building in Europe’s engineering heartland. In the physical industries, founders rarely get the chance to openly compare notes on their lessons learned, different decisions they have made and why. Whether it is about a pilot that took 18 months to convert, or a corporate deal they chose to walk away from. The ‘full circle’ moment happens when those journeys finally meet. When your approach to navigating roadblocks the hard way becomes collective knowledge for the rest of the table, closing the loop on the challenges behind building in this sector. For us as investors, the evening was a reminder of the unique realities these founders face every day. Much of our conversation centered on the immense patience required to sell to the European Mittelstand, where contracts rely on a deep track record of trust long before commercial logic ever enters the equation. That slow pace is compounded by heavy administrative paperwork and bureaucracy. Yet perhaps the most interesting dynamic was how founders build credibility in a market where their strongest proof points cannot be shared, because the commercial milestones that would speak loudest are locked behind NDAs and kept entirely off the record. The rest of the conversation stays in the room, but the circle keeps growing. After Berlin in May and Stuttgart in July, we are heading to Munich, Barcelona, Stockholm, Zurich, and London next. If this table sounds like yours, stay tuned and subscribe to our Luma calendar ⤵️ https://lnkd.in/gDZu3Cj8 👏 Thank you to every founder who joined us and to Acton Capital for co-hosting this event with us. #kompasvc #fullcircle #stuttgart cc: Sebastian Peck Ilena Mece Nick Gosen Dominik Alvermann Joseph Hefele Laserhub GmbH Resourcly Telura Deadalus Systems Solarize fruitcore robotics Assemblio Vyoma arflo Fixel Workwise Max Plank Institute for Solar System Research

Warehouses lose up to 10% of revenue to shipment inaccuracies and disputed claims. Most of that money is technically recoverable because the evidence already exists on the warehouse floor. The problem is what comes next. Resolving a single overage, shortage or damaged goods claim (OS&D) takes 12 to 15 hours of back-office work on average, and an estimated 30% of claims filed are fraudulent. Without visual evidence, operations have little choice but to pay them. Arvist was built around the conviction that the best warehouse technology fits into existing operations rather than disrupting them. Their vision AI platform turns the cameras, warehouse management systems (WMS), and tablets you already run into real-time quality control, compliance monitoring, and visual audit trails, requiring no new hardware or process overhaul. Today, that same idea extends to the claims side of the business. Arvist newly launched Claims Management product automates the full OS&D claims process from initial documentation through dispute resolution. Visual evidence captured during quality control is automatically compiled into completed claims forms, with photos, timestamps, and shipment details already attached. It connects to the more than 80 WMS and ERP platforms Arvist already integrates with and deploys in days using existing infrastructure. As Founder and CEO Nilay Parikh put it: "We built Arvist so warehouses wouldn't have to change how they work to get better results. That started with automated QC stations catching problems at the source. But claims and chargebacks still show up, that's just the nature of shipping. Claims Management is us applying the same idea to that side of the business: automate it, don't make anyone change their process to use it." In practice, that means 3PL providers, distribution centre managers, OS&D coordinators, and finance teams can stop absorbing costs they should be recovering. Claims Management handles what happens when errors do occur, so those costs get recovered rather than written off. 👏 Congratulations to Nilay Parikh and the full Arvist team on a meaningful product milestone. If you carry the claims burden today, their latest press release is worth a read ⤵️ https://lnkd.in/gmVrbfud cc: Talia Rafaeli Summit Rosenberg #kompasvc #arvist #warehousemanagement #visionai

Lessons in the physical industries are often learned the hard way, by founders who rarely get the chance to compare notes. Our Full Circle dinner series exists to share those lessons back around the table. After hosting our last ecosystem dinner in Berlin in May, we are headed to Stuttgart in July. Few cities carry more industrial weight than Stuttgart. Stuttgart is the engineering heartland of Europe, home to the Mittelstand and the automotive giants, and now to a generation of founders rethinking how the region designs, builds and moves things. The idea behind Full Circle is simple. We curate a group of founders who all share the conviction that the physical industries are ready to be transformed. Each of them have spent years putting that belief to work inside their own company. Around the table, off the record, those journeys finally get the chance to meet. The pilot that converted after eighteen months. The production ramp that nearly broke the company. The corporate deal that was right to walk away from. What a founder learned the hard way becomes collective knowledge. That is the full circle moment our dinners are built for, city by city. Each edition brings together around twenty-five founders: portfolio companies, alongside exceptional founders from the wider ecosystem, invited by referral. This Stuttgart edition is co-hosted with Acton Capital. 🌀 The circle keeps growing. If this table sounds like yours, you know what to do ⤵️ https://luma.com/rpy5vrd8 Sebastian Peck, Ilena Mece, Nick Gosen, Dominik Alvermann, Joseph Hefele

We share common ground with every founder we back. Earlier this week, they discovered how much they share with one another too, over a three-day retreat that started in Barcelona and ended in the Pyrenees as part of our inaugural Founders' Summit - "Common Ground" The idea was simple. Our founders are problem solvers at different stages of the same journey. They face similar struggles and decisions, and tackle them in different ways. They rarely have a forum to discuss their common challenges and strategies to overcome them. So we built one. Every session during these three days was curated from input received from our founders on what content would be of most value. We chose one special weekend - the Grand Depart of the le de Tour France. This iconic event symoblises to us the trials and tribulations that founders endure as they scale their companies - the uphill battles, the downhill thrills, and the twists and turns along the way that are integral to the entrepreneurial experience. We opened on July 4th with rooftop drinks overlooking the beautiful Barcelona skyline, and mingled to the backdrop of music by our very own DJ John Miller, investment analyst by day. From there, we traded the city for the mountains of the Spanish Pyrenees. The program included a rich offering of storytelling inspiration to strategies for founders as they lead, build and grow. Vika Victoria (Vika Talks) opened with a profound session on how the language we use shapes how the world sees us. Vanessa Roberts followed with a session on wellbeing, and why trust, not control, is what creates the conditions for performance. Warner Philips drew on 25 years as an entrepreneur and investor to speak openly about scaling and the road to exit. Fredrik Halmøy Wisløff from Findable gave a founder-led deep dive into how AI agents are transforming the way his team works. And Marijn Kouw brought it all down to execution, closing with a GTM breakdown and hackathon that sent teams home with actual work done. Just as much happened between the sessions. Mornings started with time for introspection, and meditation, including breathwork and yoga led by our Head of Marketing Pauline Jimenez. Down time between sessions meant casual conversations, exchanges of ideas and sparks of creativity over strolls by the river, long lunches, poolside chatter or by the evening BBQ. We closed the Summit with hiking or canyoning, including a 10m cliff jump. Founders supported each other, discovered how much they had in common - from shared challenges to real grounds for collaboration - and spoke more openly in nature than in any boardroom. That is the community we set out to build, and why "Common Ground" is only the first of many summits to come. A lot more content from those three days is on its way, so stay tuned. 👏 A special thank you to the KOMPAS team for putting together an exceptional experience, and to every founder who showed up with openness and energy. #kompasvc #commonground2026

71% of industrial leaders are using AI in their operations, yet only 6% have scaled it enterprise-wide. That gap was the story at Automate Show 2026 last week, where Ilena Mece and Yoobin Jung spent the week on the ground with the people closest to what is happening in physical operations. The conclusion from the week was that the technology is not the bottleneck to making physical operations more productive, but organisations are. Middle management accounts for an estimated 35% of AI adoption resistance, and 70 to 90% of AI pilots never scale. The blockers, when you decompose them, are cost, legacy integration, and data quality, not model performance, contrary to what one may think. Three sessions stood out to our team: First, Joshua Joseph at Tesla walked through AMR deployment at the Austin Gigafactory, where the numbers are extraordinary: 97% fewer industrial vehicle accidents and 99% fewer manual forklift moves, yet a 12% manual intervention rate persists at steady state. Resilient autonomy is a synthesis of physical infrastructure, digital intelligence, and human integration working together, and removing any one of the three breaks the system. Second, Kevin Reese from Agility Robotics and Dr. Riccardo Mariani from NVIDIA made the case that physical AI safety cannot be retrofitted. Standards governing legged robots are being written right now, and the moat that early compliance creates will not be available indefinitely. And third, Steve Minniear and Krishna Tulugu from BorgWarner and Francis Vatakencherry from Siemens showed how their digital twin started not with AI but with four years of automating energy data collection across 75 plants. Only then did predictive maintenance become possible, delivering four times the projected ROI. Their warning holds: AI applied to a broken process amplifies every inefficiency rather than resolving it. Durable companies will own the integration layer, the data infrastructure, and the change management process that converts a pilot into a production system. That layer is harder to build, harder to copy, and far stickier than another model wrapper. We also hosted a dinner alongside Arvist and fruitcore robotics with senior executives from NVIDIA, Siemens, Dell Technologies, Schaeffler, Microsoft, Prologis, and DHL and enjoyed some candid conversations about what ‘production-ready’ means to enterprise buyers amongst many other topics. ⤵️ If you would like a seat at our next dinner, reach out to Yoobin Jung or visit kompas.vc/events. If you are a founder building in physical AI or robotics, Ilena Mece would love to hear from you. #kompasvc #automate2026 Nilay Parikh Dr. Stephan F. L. Fabian Kienzler Fabio Manfreda Paul Brisson 🚚 📦 🌱 Martin Magrian

💥 Scheduling runs the physical world. Every technician dispatched, every shift rostered, every route optimised. But as AI transforms how software is built, a critical gap is emerging. While LLM-generated schedules look right, often they are not. In mission-critical operations like field service, workforce management, and logistics, almost correct is not good enough and can be costly. Correctness, compliance, and reproducibility are non-negotiable. Deterministic optimisation is not a feature but infrastructure. That is why we invested Timefold's $13M Series A alongside Alstin Capital who led this round, with continued support from Lakestar and Smartfin. Timefold gives software teams API-first scheduling optimisation. Developers embed enterprise-grade vehicle routing and shift scheduling directly into their applications. The platform handles any constraint, at any scale, with guaranteed correctness. It is built on 18 years of open-source development that powers some of the most complex scheduling environments in the world. In 2025, ARR grew 4x. NEC Software Solutions (India), CBRE, Lufthansa, Thales, and Subaru Corporation run mission-critical scheduling on Timefold today. Maarten Vandenbroucke and Geoffrey De Smet understood early that as AI generates more software, the infrastructure underneath it needs to be deterministic. That conviction built the world's most battle-tested scheduling engine. The market is now catching up. Welcome to the KOMPAS family. 🔗 Read more at Tech.eu > https://lnkd.in/eATb9HDq Thank you to Tamara Djurickovic for covering the story. Sebastian Peck Nick Gosen John Miller Alstin Capital Alexander Meyer-Scharenberg Margret Dupslaff Lakestar Enrico Mellis Smartfin Thomas Depuydt

💥 Anyone still looking for the silver bullet in decarbonisation is asking the wrong question. Last week at South Summit in Madrid, our colleague Summit Rosenberg joined Romain Diaz (Satgana) and Nina Pfifer (Spin Capital) for a panel on systemic decarbonisation. Here is what we took away. The vibe shift in climate is real, but it is being misread. LPs and founders drifting toward AI and defence are not wrong to follow economic tailwinds, but the narrative that climate tech is losing ground misses what is actually happening. The companies having their best moment right now are the ones at the intersection of climate and economic necessity: energy security, critical minerals, supply chain resilience. The reframe is not away from climate. It is toward climate solutions that do not need policy support to survive. This brings us to the green premium. For too long, the space assumed buyers would pay more for the sustainable option. At scale, they will not. Products have to win on economics. That is the bar, and it is a more durable foundation for building a real business than any subsidy or regulatory tailwind. The structural problem is the missing middle. Early-stage VC gets companies to a pilot. Infrastructure funds want proven, cash-flowing assets. The gap in between, the €5M to €15M tickets needed to go from one reference customer to commercial scale, is where too many strong companies stall. As @Nina put it during the panel: “Understand your capital needs before the capital markets can meet you, and start those conversations years before you think you need them.” Found Industries is a good example of this. Starting with industrial decarbonisation via aluminium fuel, the company expanded into critical metals recovery before gallium supply chains became a geopolitical flashpoint, resulting in a business that serves both a long-term energy thesis and an immediate national security need. That kind of adaptability is what separates the companies that reach scale from those that plateau after their first pilot. The next decade of climate progress will not be driven by the one breakthrough technology. It will come from the aggregation of many solutions, deployed across many value chains, financed by capital stacks that are far more creative than traditional VC alone. The funds and founders building for that reality are the ones worth watching. Watch the full conversation ⤵️ https://lnkd.in/e3kBtiQ7 Summit Rosenberg #kompasvc