Investment roadmap / 01
Physical systems, rewritten.
The defining machines of the next decade will not be measured by component count. They will be measured by how much useful work they extract from every sensor, actuator, watt, operator hour, and dollar of deployed capital.
A working underwriting model for the intelligence layer
The dislocation
Physical innovation is still priced as if intelligence must arrive inside expensive apparatus.
Better models, cheaper edge compute, abundant sensors, and modern control systems have changed the design space. Functions once purchased through optics, metrology, infrastructure, or mechanical precision can increasingly be produced through software. The cost structure has moved. Most markets have not caught up.
01
Intelligence becomes the component
The model, control loop, and deployment data carry an increasing share of system performance.
02
Lower cost expands the market
A radical cost change reaches customers, geographies, and workflows that legacy equipment cannot serve.
03
Deployment compounds the advantage
Every installation produces operating evidence that improves the next system without rebuilding the physical stack.
The architectural shift
We invest when the intelligence layer changes the system, not when it decorates it.
01
From
Specialized sensing
To
Inference from abundant signals
Use computation to recover precision from inexpensive cameras, microphones, inertial sensors, and optical components.
02
From
Purpose-built machines
To
Software-defined platforms
Move adaptation into perception, planning, and control so one physical architecture can learn to perform more work.
03
From
Greenfield replacement
To
Brownfield augmentation
Create retrofit paths that respect the installed base and let customers capture value before replacing infrastructure.
04
From
Lab-bound precision
To
Distributed capability
Design systems that preserve the outcome while reducing calibration, operator skill, transport, and environmental requirements.
Where we invest
The intelligence layer is the leverage.
We invest across sectors, but the company must own a technical mechanism that improves the economics of a physical outcome.
Area / 01
Robotics and autonomy
Perception, planning, manipulation, and control systems that allow useful machines to operate through uncertainty, not only inside controlled environments.
Area / 02
Industrial intelligence
Software-defined inspection, production, maintenance, and material movement that improve throughput without demanding a greenfield facility.
Area / 03
Edge systems
Compute, sensing, and inference architectures that deliver fast, resilient decisions where bandwidth, power, and infrastructure are constrained.
Area / 04
Accessible instruments
Scientific, clinical, agricultural, and cultural systems that replace scarce equipment with distributed intelligence and radically wider access.
Our underwriting filter
What earns technical conviction.
We underwrite
- A first-principles insight that changes the cost-capability curve
- Defensibility in architecture, control, data, or operating workflow
- A credible path from prototype to repeated, margin-accretive deployment
- A market that expands materially when system cost falls
- Founders fluent in both the equation and the customer environment
We challenge
- Component novelty presented as customer value
- Products that require permanent founder-led installation
- Laboratory precision with no field-learning architecture
- Unit economics that appear only at implausible manufacturing scale
- Automation that improves a task without changing the buyer's economics
The founder question