Tag
15 posts
A new speech model promises hour-long transcriptions in seconds, entirely on your Mac. But the most eye-catching comparison depends on benchmarks that may not be measuring what they seem.
The next PC upgrade may be less about speed and more about who—or what—gets to use your files. Microsoft’s pitch promises convenience, but its biggest test may be earning trust while agents act on your behalf.
A viral video proves a 2020 Apple Watch can run a powerful local LLM. So why are tech giants still selling you on the cloud and expensive new hardware for AI?
Running a 35-billion parameter AI model on a phone sounds impossible, given it needs 20GB of RAM. But a new technique flips the script on memory limits, opening the door for powerful, private AI that never leaves your device.
Everyone assumes on-device AI requires the latest hardware, but they're wrong. A 6-year-old Apple Watch just proved that powerful, local LLMs are already possible on the tech you own.
Everyone thinks on-device AI is limited to tiny models, but a new technique just shattered that assumption. The breakthrough isn't a new chip, but a clever memory hack that Apple's OS was already secretly perfect for.
Most AI development is stuck in a costly cloud-based arms race. A quiet European lab is proving that the future of AI isn't in the data center, but right on your device.
Google just launched LiteRT.js, a new library that brings near-native AI speed directly into your browser. Here’s why its WebAssembly core makes TensorFlow.js obsolete and unlocks a new era of on-device machine learning.
A massive 27 billion parameter AI now runs entirely on your phone, with no cloud or Wi-Fi needed. This breakthrough in model compression delivers unparalleled privacy and performance, right in your pocket.
Two weeks ago we mapped what iOS 27 would let developers build. Apple gave the WWDC 2026 keynote on June 8 — so we graded ourselves. Both big bets landed: the Foundation Models framework now takes image input (the 1–2GB bundled-VLM tax is over), and App Intents became the way into Siri (with SiriKit now deprecated on a 2–3 year clock). We even called the Core AI rename. What we missed: Apple outsourced Siri's brain to Gemini (~$1B/year) and made MCP system-wide across iOS 27 and macOS 27. The reality map for what to build now that the rumors are facts.
WWDC 2026's keynote is June 8, and most of the iOS 27 rumors are consumer polish that won't change your roadmap. Two will: multimodal on-device Foundation Models (which could turn the 1–2GB vision models apps bundle by hand into a system API) and Siri Extensions + App Intents (free distribution you can wire up today). A builder's opportunity map, a day-one watch checklist, and what to ship this week.
Tired of expensive, slow, and privacy-invading cloud TTS APIs? A new local model called Supertonic 3 runs entirely offline on a CPU, and it's surprisingly good.
Google quietly released an app that turns your phone into a powerful, private AI that works completely offline. Here's why it's a game-changer and how you can start using it today.
Google just released Gemma 4, a family of open-source AI models so efficient they can run powerful, agentic workflows directly on your phone or laptop. This changes everything for developers, privacy, and the future of AI.
CES 2026 is no longer about far-off promises and flashy concepts. This year, 'physical AI' takes center stage as real-world robots and on-device intelligence move from labs into our factories and homes.
