Intelligence

Apple platform.
Artificial intelligence.
The convergence.

Edge AI is Apple's thesis. On-device inference, the Neural Engine, Private Cloud Compute, the Foundation Models API — a platform bet that intelligence belongs on the device, governed by the user, not the cloud. This section tracks the technology, the governance architecture underneath it, and what the practitioner needs to build on it.

The Classical Programme

Classical education
for the age of artificial intelligence.

“Technology alone is not enough — it's technology married with liberal arts, married with the humanities, that yields us the results that make our heart sing.”

— Steve Jobs

The seven liberal arts mapped to Apple's development stack. Euclid's axioms through manifold geometry, Grapher through gradient descent, formal methods through SwiftVector. A unified curriculum from ancient foundations to governed intelligence — built on the conviction that the deepest practitioners are the broadest thinkers.

Three goals:

  • Foundations in mathematics and formal reasoning
  • Certification preparation — Apple, cloud, platform
  • Masters in AI prerequisites — CU Boulder MS CS baseline
Begin →
Three tracks
Practitioner

Foundations. Euclid to embeddings. Grapher to gradients. The mathematics that makes AI legible, mapped through Apple's own tools. Where the curriculum begins.

Architect

Systems. TLA+ to SwiftVector. Formal methods applied to governed intelligence. The composition problem solved at the type-system level.

Researcher

Theory. The manifold hypothesis. Attention as geodesic routing. Publication-grade depth for practitioners who want to understand, not just use.

Three streams

Technology · Governance · Practice
On the technology

Apple's AI architecture, where it's going, what the practitioner sees that the journalist doesn't. The edge thesis. The Gemini deal. FoundationModels as ecosystem play. Private Cloud Compute as the only cryptographically sound answer.

On the governance

The Agency Paradox and its descendants. ClawLaw. The constitutional layer. What governed autonomy actually looks like implemented on Apple Silicon. The legal and regulatory dimension as it develops.

On the practice

What it means to write with AI now. ChronicleLaw as a case study. The collaboration model. Where attribution belongs. What human intelligence still does that the machine can't.

Featured & Dispatches

Position papers · Current developments
The Agency Paradox: Governed Autonomy as Infrastructure

DevOps unified software creation and software operation under one discipline. Governed autonomy seeks to unify AI reasoning and operational authority under one control-plane model. The pattern is proven. The application to autonomous AI agents is the work remaining.

Read the essay → ClawLaw implementation ↗
Law at the edge: why AI governance belongs on Apple Silicon

The case for on-device constitutional governance and why Apple's architecture makes SwiftVector the reference kernel.

The composition problem: why per-action governance isn't enough

Individually compliant actions that collectively constitute scope creep. Why AI governance needs session-level state.

The Gemini Deal Is Not a Defeat — It's a Pivot Toward Shipping

Apple partnering with Google on Gemini reads as capitulation if you don't understand the architecture. It reads as pragmatism if you do. The on-device layer is the moat. The cloud layer is the bridge.

What Rockwell Knows That Giannandrea Didn't

Apple replaced its AI chief. The new appointment tells you everything about what Apple thinks went wrong — and what it plans to do differently.

Three Lines of Swift: Why FoundationModels Is the Sleeper Story of WWDC 2025

Three lines of Swift to run a foundation model on-device, privately, with no API key. Apple buried the most important announcement in a session about developer tools.

Private Cloud Compute Is the Most Important AI Architecture No One Is Talking About

Not policy. Cryptographic design. Apple built the only cloud AI architecture where the operator provably cannot read your data. Security researchers can verify it because Apple publishes the software for inspection.

The Admonishment: On Being Told by an AI to Document Your Results

The AI models have admonished the author more than once to document results. That admonishment is itself a finding — and a proof point for the governance argument.

For hands-on tutorials working with Xcode's Apple Intelligence integration:

Developer tutorials →