210 Problems We Have Learned to Call Normal

These problems no longer look like problems. They look like the unavoidable price of building useful AI, machines, instruments, electronics, and infrastructure. That is exactly why I am publishing all 210 at once.

I plan to create three focused subsidiaries, each attacking a different layer: capability-native agency; live reconfigurable physical computation; and minimal-apparatus physical intelligence.

In this plan, the same customer industry may appear under more than one subsidiary because each attacks a different layer. A robotics company, for example, can face an authority problem, a computational-structure problem, and an interface-apparatus problem at the same time.

Read every bullet as beginning with “Of course…” If a sentence feels obvious, ask when and why we accepted the burden it describes as inevitable.

1. Capability-native agency

I plan to build the first subsidiary around systems in which AI agents and autonomous machines possess structurally bounded authority, rather than ambient power constrained by monitoring, policies, and retrospective accountability.

Segment 1A: Enterprise AI agents acting across production systems

Segment 1B: Multi-agent systems and delegated machine organizations

Segment 1C: Autonomous industrial and robotic action

Segment 1D: Capability-enforced computing infrastructure

2. Live reconfigurable physical computation

I plan to build the second around spatial, locally owned, dynamically reconfigurable computation for instruments, robots, satellites, industrial systems, adaptive edge machines, and eventually programmable matter.

Segment 2A: Scientific and technical instruments

Segment 2B: Industrial systems and robots

Segment 2C: Satellites and remote autonomous systems

Segment 2D: Adaptive edge machines

Segment 2E: Programmable matter and large spatial electronics

3. Minimal-apparatus physical intelligence

I plan to build the third around near-sensor computation, direct physical interfaces, tiny local learning, unusual active devices, and systems that remove converters, centralized machinery, or inaccessible fabrication where those layers constitute the real burden.

Segment 3A: Near-sensor computation and distributed sensing

Segment 3B: Direct physical interfaces and converter-heavy systems

Segment 3C: Tiny local learning and adaptation

Segment 3D: Unusual active devices and accessible fabrication