Blog
Exploring AI, engineering, and the future of technology
Filter by Topic
Latest Stories
- engineering 5 min read
The Command You Approve Is Not the Command That Runs
Why approved commands run as something else: shell expansion rewrites what your agent's safety check read. A case for guarding at execution.
swift ai-agents security - engineering 12 min read
Wax: One File That Survives Hard Kills and Context Resets
I built Wax, a single-file crash-safe memory engine for AI agents on Apple Silicon: WAL durability, hybrid RAG, bitemporal facts, no cloud dependency.
imported - engineering 21 min read
Inside Swarm's WebSearchTool: How I Built It to Fit a 4K Context Window
I built Swarm's Swift web tool for 4,096-token on-device LLMs. Here's the tiered envelope, the Wax-without-embeddings bet, and fourteen decisions behind it.
imported - engineering 11 min read
How I Conducted Deep Research with a 3B On-Device Model and a 4K Context Window
A 3B on-device model with a 4K context window produced a 2,336-word grounded research report using live web search. Here's the architecture that made it work.
imported - engineering 6 min read
EdgeRunner Beats llama.cpp: A Swift/Metal LLM Engine on Apple Silicon
Built EdgeRunner in pure Swift and Metal from scratch over a weekend. On Qwen3-0.6B Q8_0, it hits 212 tok/s on M3 Max, beating llama.cpp by 16%. Here's what w
imported - engineering 10 min read
What Our Espresso Throughput Experiments Actually Found
A throughput research postmortem from Espresso: what we tested, what failed, what stayed slower than baseline, and where the real architectural blockers are.
imported - engineering 5 min read
Why iOS Developers Need Conduit: A Unified SDK for Every LLM
If you've shipped AI features in an iOS app recently, you know the drill. Each provider has its own SDK. Switching means rewriting everything. Conduit fixes t
imported - engineering 10 min read
ContextCore: GPU-Accelerated Context Memory for On-Device AI Agents
We built ContextCore to manage conversation context on Apple Silicon without cloud or battery drain. Four-tier memory, Metal shaders, 63M chunks/sec, sub-5ms
imported - engineering 7 min read
EdgeRunner Architecture: What We Found When We Read the Whole Damn Codebase
We fully documented EdgeRunner, a Metal LLM inference engine for Apple Silicon. Here's what we found that was actually interesting.
imported - engineering 8 min read
Why On-Device Observability Is Non-Negotiable for GenAI Applications
How local telemetry transforms AI reliability, privacy, and performance for applications running on edge devices.
imported - engineering 7 min read
Colony: A Native AI Agent Runtime for Apple Platforms
Colony is a Swift framework that orchestrates AI agent loops on iOS and macOS using Apple's Foundation Models. Here's why it exists and how it works.
imported - engineering 7 min read
Swift for Coding Agents
After building a multi-agent framework in Swift, I have thoughts on why it might actually be a better fit than Python or TypeScript for this kind of work.
imported - Prompt Engineering 9 min read
Teaching Machines to Think: The Prompting Revolution in AI
How chain-of-thought prompting and simple techniques transformed large language models from pattern-matchers into genuine problem-solvers.
Prompting Chain-of-Thought AI Reasoning Self-Consistency - LLMs & Models 9 min read
Are We Building Conscious AI Already?
Exploring the uncertainty around AI consciousness, ethics implications, and what researchers at Anthropic are learning about model welfare.
AI Ethics Consciousness Model Welfare Anthropic - engineering 8 min read
The file system is becoming the interface for agents
Why the serious agent stacks are converging on files as the primitive for machine behavior, and what that means for the future of AI systems.
imported
Explore Topics
Deep dives into specific areas
Browse posts organized by topic to find exactly what you're looking for
View All Topics