0
On the Wire now
Free
Applied AISonos’ New Voice Assistant Is Fine for Music but Bad at Spelling (and More)Sonos 27voice being good at speaker control but weak on basic reasoning is a reminder that narrow assistants can still win if they nail a single job. Hardware teams don’t need frontier models everywhere — they need voice stacks tuned to their domain, with clear guardrails on what the assistant will and won’t attempt.GizmodoApplied AITwilio's CMO is restructuring the marketing department around AI 'orchestrators'Marketing org design is starting to assume AI as the default executor and humans as cross-functional conductors. If you still have siloed channel specialists, expect pressure to consolidate into fewer, higher-leverage roles that own prompts, data, and outcomes across the stack.Business InsiderStartups & VentureMasayoshi Son says passing on investing in Anthropic in 2023 was a "mistake", noting that "by that point, we had already decided to invest in OpenAI"Top-tier capital is now treating frontier labs as a portfolio construction problem, not a single-bet conviction call. If you’re building on one ecosystem, assume your investors are quietly underwriting exposure to its rivals too.Financial TimesApplied AIKPMG uses 100+ AI agents to conduct a parallel audit of a FTSE 250 company in a pilot, saying the AI asked "better questions" and dug deeper than human staffWhen a Big Four firm runs 100+ AI agents on a live FTSE 250 audit and says the system asked “better questions,” the bar for what counts as reasonable diligence just moved. If you run finance, risk, or internal audit, start scoping where agentic reviews can augment sampling and anomaly detection before regulators and boards start asking why you aren’t.Financial TimesStartups & VentureBuy the AI boom or beware the hype? 5 investing pros weigh in on Anthropic’s IPOAnthropic’s IPO is as much a sentiment gauge as a capital event—how public markets price it will feed directly into late-stage AI valuations and partnership leverage. Operators should watch the multiple and lockup behavior to calibrate how much equity-driven aggression their AI vendors will be under in 2027 budgets.Business InsiderApplied AITencent Says US Chip Curbs to Slow Chinese AI in the Short RunTencent openly framing US data center curbs as a short-term drag and a long-term push toward domestic semis means China AI timelines are stretching, not stopping. If you sell into Chinese AI builders, assume a bumpy 12–24 months and start mapping to local chip ecosystems and onshore partners now.Bloomberg Technology

Latest Weekly Signal & Field Reports

The Price of DoneField Report

The Price of Done

Model Signals field report by Isaiah "Zay" Steinfeld (Neue Alchemy) on Claude Opus 5.5 and GPT-6 Sol/Luna: cost per successful agent task fell about 22x in five months, but most agent failures are still reported as…

Sep 22, 2026
The Loop Closes — Field Report · SpaceX × CursorField Report

The Loop Closes — Field Report · SpaceX × Cursor

SpaceX didn’t pay $60 billion for a code editor. It bought the system around the work — compute, models, evaluation machinery, work surface, distribution — and the filings show it was structured that way from the…

Aug 14, 2026
The Checkpoint Travels. The Fingers Don’t — Yet.Field Report

The Checkpoint Travels. The Fingers Don’t — Yet.

Google DeepMind didn’t ship a robot. It shipped the layer above every robot — and the arbiter between the thinking and the moving. One checkpoint now drives three bodies; the gap in what it can’t do is the whole report.

Jul 30, 2026