Deep-tech doesn't fail at the technology. It stalls when complexity compounds faster than the coordination and trust built to absorb it. CompoundWorks exists for that exact moment.
This is for you if:
Free, 10 minutes — see exactly which dimension is holding you back.
The Scaling Readiness Check locates your venture across Technology, Organisation and Trust — and names the one dimension holding the other two back.
Take the Check →Why proven technology stalls between pilot and platform, and what breaks at the second integration.
The Compound Innovation Gap →An executive diagnostic session on your situation. Credited in full toward a sprint if you go further.
Book a session →Some technologies grow in parallel. Others start to compound — and Physical AI, industrial systems, and defence & dual-use are exactly where that’s happening right now.
Layers stop adding up and start amplifying each other. That interdependence is the source of the upside — and of the difficulty.
Where this is headingThe distance between a system that works under controlled conditions and one that absorbs real multi-vendor, multi-site complexity.
Read the paperThe technology works and the first customers are real, but the business stops compounding. Five patterns account for most of it.
See the five failure patternsDeep-tech ventures rarely stall because nothing is improving. They stall because improvement is uneven — technology outruns the organisation, or commercial traction outpaces trust. The Scaling System Maturity Framework (SSMF) exists to make that imbalance visible: it tracks maturity across three coupled dimensions — Technology, Organisation and Trust — that must evolve together, over four levels of maturity. Each level is a qualitative shift in how the venture behaves under pressure — how much complexity it can absorb without losing coherence and cadence — not an incremental improvement. The full framework lives here. Here is what the Check — the SSMF applied to your venture — actually returns.
Trust sits one level behind the other two. Delivery is reliable because specific people make it reliable — which reads internally as strength and externally as opacity.
What breaks next: the second institutional buyer asks you to evidence reliability without a relationship, and there is nothing to hand them.
Illustrative profile — not a real client.
Everything starts with the Check. What follows is defined by what it uncovers — not a package chosen in advance.
Technology lagging → the Architect. A direction question: R&D roadmap, platform architecture, the runway built before a customer exposes its absence.
Organisation or Trust lagging → the Operator. A delivery question: decision rights, earned trust, the crossing from managed growth to repeatable delivery.
Either path, followed honestly, usually ends up needing both.
See how the two routes work, in fullSystems rigor, platform thinking, and business execution — together, not separately.
Jose Lopez founded CompoundWorks after 25 years building and scaling deep-tech companies across defence, aerospace, mobility, industrial systems, XR, simulation and digital twins — solving mission-critical problems in regulated environments, then turning what repeated into platforms, new categories, and enterprise adoption.
The practice reads that accumulated experience two ways. The Architect spotted convergence before the market named it — digital twins, the industrial metaverse and Physical AI — delivered mission-critical and simulation systems, extending simulation's use from training into full-life-cycle support for military systems and operations, and built new categories rather than waiting for them to arrive. The Operator has done more than run those systems inside demanding organisations: building multidisciplinary engineering organisations and maturing them into high-performing teams, delivering under pressure rather than just designing for it.
Narrow is replaceable — a specialist can be hired for any single piece of this. Compounded across three convergence points and 25 years is not. CompoundWorks is the evidence that compound works.
Defence & aerospace · Industrial systems & mobility · XR & spatial computing · Simulation & digital twins · Energy & critical infrastructure
Grew NADS, a profitable, €5M-revenue defence engineering business with 70+ people, as a trusted partner in the military supply chain, then led its international expansion through the deep-tech spin-off Simware Solutions — reaching customers across Europe, the US and China
Co-founded TMRW, helping scale it to 150+ people and €10M+ in funding, while personally building and leading a 50+ person team across Europe, India and the GCC, and contributing to 10+ patents in digital twins, AI and spatial computing
Delivered with and for organisations including Spanish MoD, NATO, Airbus, Thales, Lockheed Martin, BMW, Porsche, Daimler Truck
Not just a visionary technologist — but an innovator with a rare capacity to navigate the complexity of deep-tech products and organisations.
He sees the big picture and the smallest implementation details simultaneously — and solves complex problems quickly while thinking well outside the box.
He translates complex technical concepts into actionable strategies that drive projects to success — regardless of the challenges.
Nothing here is hidden behind a door. Every page is grouped by the question it answers.
Posts and short essays as the practice develops — the same arguments worked out in public, before they’re settled enough for this site.
The visible problem is rarely the real constraint. Diagnosis has to come before intervention — until you find where the system is losing compounding power, even good people end up solving the wrong thing.
Read on LinkedInMost teams using AI agents feel faster, but throughput hasn't improved. That's not an AI problem — it's a system maturity problem, and it's exactly where Hybrid Agile gets more relevant, not less.
Read on LinkedInValue emerges from integrating sensing, spatial abstraction, cognition, compute and governance into one coherent stack. Most failures don't happen inside the model — they happen at the interfaces.
Read on LinkedInTen minutes will tell you more than another quarter of pushing harder on the one that isn’t.