Flyability SA: Reinventing Industrial Inspection with Indoor Drones

The Elios 3 is Flyability’s indoor inspection drone, designed for data collection and mapping in complex, confined and inaccessible industrial environments. Its FlyAware™ engine combines computer vision, LiDAR and SLAM-based stabilization to create 3D maps in real time. Its collision-tolerant design enables operation in confined spaces and access through openings as small as 50 x 50 cm. It features a 4K camera, thermal camera, distance sensor and powerful lighting. A modular payload bay supports applications such as ultrasonic thickness measurements and radiation detection, while a flight time of up to 12.5 minutes supports inspections of industrial assets.

Interview questions for Patrick Thevoz, CEO & Co-Founder, Flyability SA:

«Flyability was founded in 2014 with a clear mission: keep people out of dangerous spaces. What was the original spark behind this idea — and how closely does today's company resemble what you first envisioned?»

Two events, a year apart: the Haiti earthquake in 2010 and Fukushima in 2011. In both cases the world had rescue and inspection robots, and in both cases they were useless as they couldn't get into the rubble or into the reactor buildings. The bottleneck wasn't sensing or intelligence, it was the physical ability to move through a cluttered, unmapped, GPS-denied space without destroying yourself. Adrien had been studying how insects fly: a fly hits a window, tumbles, keeps going. That became Flyability.
The mission hasn't moved. Two things around it have. We aimed first at disaster response and found that catastrophes are rare while industrial confined spaces are everywhere and inspected on a schedule. Same problem, viable business. And we went from selling a tool to delivering a solution: the drone is now one component alongside payloads, software, data and training. Customers don't buy a flying machine, they buy an inspection solution that no longer requires a human inside the asset.

«The Elios 3 is described as the world's first collision-tolerant drone. You've said it's easier to operate than most toy drones — yet it's been deployed in nuclear reactors and Alpine glaciers. How do you achieve that balance between simplicity and extreme performance?»

I'd reframe it slightly: the Elios isn't simple, it's forgiving. A consumer drone is easy because it has GPS, open sky, and avoidance that tells it to stop. None of that exists inside a boiler or a mine shaft. So instead of avoiding contact we designed around it with a protective cage decoupled from the flight system, so the drone can touch a wall, absorb it, and stay stable. Once a collision is no longer catastrophic, the pilot stops flying defensively. That's where the ease comes from. A significant part of our R&D is spent on ease-of-use features, using AI to make the operator forget he is piloting, just acquiring data in his asset.

But I'd add something people miss: flying the drone was never the hard part of this job. The hard part is capturing the right data and interpreting it correctly: knowing which weld to look at, what a millimetre of wall loss means, whether that crack is new. Our customers are boiler engineers and inspectors, not pilots. Our job is to make the flying disappear so their expertise is spent where it's actually worth something.

«Flyability has won over 11 technology and business prizes and now serves nearly 1,000 clients in over 60 countries. Which achievement are you most proud of — and which moment felt like the real turning point for the company?»

Prizes are pleasant. The number I actually watch is that we now have over 3,000 drones deployed and, more importantly, the usage rate of those drones is climbing sharply month over month. That second figure is the one that matters, because a drone on a shelf means we sold a gadget; a drone flown every week means the customer built it into how they work. Rising utilisation is the only honest proof that our customers are succeeding, not just that our sales team is.

And then there's the part that doesn't appear in any dashboard. Every so often a customer tells us a life was saved, or an industrial catastrophe was caught before it happened, because someone flew instead of climbing in. That's the real scoreboard. Confined space work kills people every year, reliably, in every industrialised country. Each entry that doesn't happen is an achievement.

«Flyability operates across power generation, oil & gas, mining, maritime, wastewater, and chemicals. Which industry has surprised you most in terms of adoption — and where do you see the next big wave of demand coming from?»

Mining, by a distance. It was a small market for us at the start, and over the last three years it has become one of our largest. What changed is that we stopped being purely an inspection tool there and became part of the operation itself: assessing the situation underground after blasting, for instance, so the team knows what they're walking into before anyone walks in. That's a daily operational workflow, not an annual inspection event, and the usage profile is completely different.

Nuclear was the other surprise, for the opposite reason. On paper it's the worst early market: slow, conservative, allergic to novelty. But when your alternative costs radiation dose, a drone stops looking experimental very fast. As for the next wave, it isn't a new market. It's the scale-up of the successes we already have. We can multiply this company by ten in the industries we're already in. Chasing novelty would be the mistake.

«With clients in over 60 countries, a network of 50+ distributors worldwide, and offices in Switzerland, the US, China, and Singapore — Flyability is by all accounts a true Swiss export success story. What have been the biggest hurdles in scaling internationally, and how did you overcome them?»

First, distance from the customer. Our buyers are engineers on plants that run 24/7, and they buy after someone credible shows them the drone inside their asset. That doesn't scale from Lausanne, which is why we built a distributor network rather than a direct salesforce in sixty countries — you need someone local who already has the plant manager's mobile number.

Second, language and culture. Technical trust is built in someone's own language, and an inspection report that a Swiss engineer finds self-evident can land very differently elsewhere. That's a real cost, and it's mostly invisible until you get it wrong. Third, support. Our drones fly in genuinely remote places such as a mine in the Andes or an offshore rig and support has to be both fast and high quality there, not one or the other. And increasingly, regulation. Drone rules, import rules, data rules: the compliance surface for a hardware exporter has grown considerably, and it now shapes where and how fast we can move.

«The confined space drone inspection market is growing rapidly. How does Flyability stay ahead of emerging competitors — and do you see the bigger threat coming from other drone manufacturers or from traditional inspection methods that are also evolving?»

Traditional methods, without much hesitation. People climbing on ropes or scaffoldings, or simply not inspecting some areas too hard to access: that's the real competitive set. The status quo is always the largest competitor in industrial markets, because it's already approved, already budgeted, and nobody gets fired for repeating last year's procedure.

Other manufacturers are a real but different pressure. Building a caged drone that flies isn't the hard part. Building one a nuclear operator will accept, with survey-grade data, a payload ecosystem, a trained partner network and a decade of case files behind it, is. Where I don't allow complacency: winning now requires both halves. Best-in-class hardware is the entry ticket but on top of it you need a reliable, secure software layer that actually delivers the data the customer needs, in a form they can trust and act on. Neither half wins alone anymore, and that's where the majority of our engineering now goes.

«Selling highly specialized hardware into regulated industries like nuclear power or oil & gas comes with enormous compliance, trust, and sales cycle challenges. What has been the hardest market or sector to crack — and what did you learn from it?»

The universal hurdle isn't a sector, it's a belief: everyone is convinced they're different. Even in use cases where we've proven the value many times over, sometimes at another plant of the same company, we're asked for a fresh proof of concept and a fresh demonstration. You learn to stop resenting it and to industrialise it instead, because a demo is often the fastest route through an approval chain that would otherwise take years. Every market has its hurdles, but maritime is particularly hard, for a structural reason: the number of stakeholders. The inspection company, the ship owner, the shipping company, the classification society, each with different interests, and no one of them able to decide alone. Worst of all, the buyer is frequently not the party who captures the savings. When the economics and the decision sit in different pockets, nothing moves quickly, however good the technology is.

«Flyability is at the intersection of hardware and deep software. How is artificial intelligence shaping the future of your inspection solutions, and what capabilities are you building for the next generation of products?»

We're investing heavily in physical AI: intelligence that acts in the real world rather than just processes text. Concretely, the Elios 3 can already fly a previous inspection autonomously and repeat it, so you get comparable data over time on the same asset rather than two flights that can't be meaningfully compared. It can also find its own shortest path home. Those sound modest; they're the foundation of everything after.

The second track is what happens to the data. A single flight produces a dense point cloud and thousands of images, and today a human expert reviews all of it — that's where the cost and the inconsistency sit. Automating defect detection, and above all change detection between inspections, turns an inspection from a snapshot into a trend line. One honest limit: in a nuclear or offshore context, an AI that flags a defect is useful. An AI trusted to say there isn't one is a much longer road, and we say so.

«"Swiss Made" carries a strong reputation for quality and precision globally. Does being a Swiss company open doors internationally, or does it sometimes create the perception of being too expensive or niche?»

We sell a premium product, and we're comfortable saying so. What our customers buy is reliability: a drone that comes back out of a reactor or a mine shaft, every time. Swiss Made reinforces exactly that positioning. It's not decoration; it's consistent with the promise. In robotics specifically, the picture is better than the old "expensive Swiss niche" cliché suggests. Switzerland has become a serious hub for AI and robotics, so being from here isn't unusual anymore, it's an advantage. It helps with customers, and just as much with hiring engineers and raising money, because investors and candidates now recognise the ecosystem rather than needing it explained.

«Flyability is a member of Switzerland Global Enterprise. What concrete value has the S-GE membership and network brought — and what would you say to other Swiss companies considering joining?»

Two distinct phases. For most of our history, we worked with S-GE when opening offices abroad: building a first network in a market where we knew no one, and getting practical help setting up. More recently the value has shifted somewhere I didn't anticipate. International shipping has become drastically more complicated: tariffs, country-of-origin restrictions, export controls that change with the political weather. For a Swiss hardware company shipping to sixty-odd countries, that's not a side issue, it's an operational risk. S-GE has been a genuine help navigating it: a place to get a reliable read on a fast-moving, much less stable world. And for Swiss deep tech, exporting isn't optional as the domestic market will never be big enough. The only question is how fast and how well you leave it.