The NeutronTech story.
The work converged.
NeutronTech began with a technical conviction, tested it in shipped software, and used the evidence to choose one commercial focus: Hubyn.
Foundation before product.
The first decision was architectural: private intelligence should run on the Apple hardware in front of the person using it. We built a native runtime in Swift and Metal 4 to control memory, scheduling, and model execution directly.
One backend: Metal 4. No translation layer. The call you make is the command the GPU executes.
Zero third-party dependencies. The runtime is fully custom and proprietary — which means every limitation we hit is ours to fix.
Research produced evidence.
We applied that foundation to bounded problems: documents, local agents, finance, and a clinical workflow. The work was not a search for unrelated markets. It tested whether the same local-first foundation could support useful, demanding software.
Early rendering, memory, and scheduling research became the native runtime for local intelligence on Apple Silicon. Neutron Engine is now an internal AI foundation, not a gaming or customer-facing product.
Document work shaped Neutron Studio's source-grounded workflow. Clinical triage became Neutron Health.
The evidence narrowed the company.
Hubyn is the commercial focus: private communication on iPhone that becomes a full collaborative workspace with local intelligence on Mac. Neutron Health is shipped proof that the capability works. Neutron Studio, Neutron Engine, and Tally are internal R&D with a defined role behind Hubyn.
Hubyn is the product, the roadmap, and the commercial focus. It is currently in private beta.
Neutron Health is shipped proof of capability. Neutron Studio, Neutron Engine, and Tally are internal R&D that strengthen the flagship.
The architecture is the strategy.
Your files, conversations, and thinking should remain under your control. For us, local intelligence is not a privacy setting added later; it is the architecture beneath the product. That choice shapes what we build, how it works offline, and where the company invests.
Hubyn's Mac intelligence and our internal Neutron Studio research run on the Mac in front of you. Only explicitly requested network work goes out.
We use internal R&D to answer concrete product questions. The result is one direction, supported by evidence rather than a collection of competing bets.
Touch the sun.
Our native Apple Silicon foundation still powers every experiment inside the lab. One mission: make Apple silicon touch the surface of the sun without melting the machine. One mandate: to get nasty about memory, scheduling, and what the GPU can actually do.
The work has a reason.
NeutronTech is built around a simple conviction: powerful intelligence should serve people without taking ownership of their work, their data, or their thinking. MagnetarMission is where that conviction becomes service — building for people working beyond reliable infrastructure and beyond the reach of the cloud.
Why we build private, offline-capable intelligence directly for Apple hardware.
Why the neutron star — and what endurance, density, and order mean to us.
How the work extends beyond products to serve people technology too often leaves behind.