Magic AI raises £8m to take its AI-powered fitness mirror to the US
THE SO WHAT
Computer-vision fitness is moving from apps and wearables into dedicated hardware — a £8m raise to push a mirror into the US is a bet that form factor and in-room presence matter more than yet another coaching app. If you’re in connected fitness or at-home health, assume the competitive set now includes AI-native hardware that owns both data and daily habit.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIYouTube releases new AI features for creators within its Studio app
When YouTube bakes idea generation and thumbnail performance analysis directly into Studio, it compresses the tooling stack around creators and pulls AI value capture in-house. If you build for the creator economy, assume more of the “optimize and ideate” surface is going to be owned by platforms, and focus on deeper workflow or cross-platform leverage.
Applied AIUpwind acquires Aegis as cybersecurity firm builds new AI Security Labs
Rolling Aegis into Upwind and standing up an AI Security Lab is a bet that “AI security” will be a distinct product and research category, not just a feature in existing tools. If you’re shipping AI into production, expect vendors to start selling dedicated AI posture and threat products — and budget a line item rather than assuming your current stack covers it.
Applied AIAustralia’s Corporate Watchdog Sets New Safeguards on AI Trading
Australia tightening rules on AI-enabled trading to match global standards is another step toward regulated, not freewheeling, algorithmic markets. If you run quant or execution systems, assume disclosure, testing, and control requirements will ratchet up — start documenting models and guardrails now rather than waiting for a supervisory letter.
Applied AIWhy effective AI communication starts with the recipient
User fatigue with obviously LLM-written content is a hard constraint — generic AI copy is now a brand liability, not a free productivity win. If you deploy AI for outbound communication, tighten prompts, add human editing, and measure response rates; volume without trust is wasted spend.