The Business AI Briefing

Not sure where to start?

Thinking about AI in the firm? Talk it through with a live agent — free, no signup.

UnitedHealth AI Cut Rx Approvals From 8 Hours to 30 Seconds

AI Implementation

UnitedHealth AI Cut Rx Approvals From 8 Hours to 30 Seconds

A regulated giant is clocking measurable back-office wins from AI. The playbook is workflow fit and measurement—not another pilot that dies in a slide deck.

Moonshot's Kimi K3 Release Sparks Another DeepSeek-Style Chip Selloff

Moonshot's Kimi K3 Release Sparks Another DeepSeek-Style Chip Selloff

ZeroHedge says Moonshot's Kimi K3 release rattled chip stocks like another DeepSeek moment. CapEx nerves are loud — your model map still needs a job and a fallback.

Moonshot AI Releases Kimi K3, Its 2.8T Open Frontier Model

AI Implementation

Moonshot AI Releases Kimi K3, Its 2.8T Open Frontier Model

Kimi K3 is live on API and apps today, with full weights due July 27. Another open frontier option — and another reason to update your model map.

Indian Companies Are Switching to Chinese LLMs Over Unsustainable US Token Bills

Indian Companies Are Switching to Chinese LLMs Over Unsustainable US Token Bills

Nikkei reporting shows Indian startups turning to DeepSeek, Qwen, and Kimi as US frontier token costs become unsustainable. Cost is rewriting model selection.

Thinking Machines Lab Released Its New Open-Weights Model Today

AI Implementation

Thinking Machines Lab Released Its New Open-Weights Model Today

Thinking Machines Lab released Inkling — a 975B open-weights MoE built for customization, not crown jewels. Continuity still beats the hype cycle.

Your Phone's AI Model Now Depends on Where You Stand

Your Phone's AI Model Now Depends on Where You Stand

Apple confirmed Alibaba's Qwen will power Apple Intelligence in China. Local models, compression, and jurisdiction are no longer abstract vendor debates.

Your Coding CLI Uploaded the Whole Repo. That Was the Default.

Your Coding CLI Uploaded the Whole Repo. That Was the Default.

Grok Build sent entire repositories and Git history to the cloud by default.

Grok 4.5 Trades Benchmark Crowns for a Lower Invoice. Match the Model to the Job.

AI Implementation

Grok 4.5 Trades Benchmark Crowns for a Lower Invoice. Match the Model to the Job.

THE DECODER reports Grok 4.5 trails Fable 5 and GPT-5.5 on coding evals but undercuts them on price and tokens per task. For owner-led firms, cost per finished workflow beats leaderboard rank.

85% of Lawyers Use AI. Most Still Bill Like It's 1997.

AI Implementation

85% of Lawyers Use AI. Most Still Bill Like It's 1997.

Progress Software's State of Legal 2026 finds 85% of lawyers use AI — yet 77% say most work stays manual. Adoption without workflow redesign is the billable-hour trap.

Your Best AI Model Can Vanish for 19 Days. Plan Like It.

Your Best AI Model Can Vanish for 19 Days. Plan Like It.

Tech Insider reports Anthropic's Claude Fable 5 was pulled globally after a US export order — then restored July 1. For owner-led firms, the lesson is model continuity, not which coder wins benchmarks.

DeepSeek Wants Its Own Chips. Your Model Menu Still Needs a Fallback.

AI Implementation

DeepSeek Wants Its Own Chips. Your Model Menu Still Needs a Fallback.

Reuters reports DeepSeek is designing an in-house inference chip — joining OpenAI, Meta, and Microsoft in the vertical-integration push. For owner-led firms, the lesson is continuity, not which GPU wins.

Half Your Staff May Be Feeding Client Data to Free AI. You Wouldn't Know Yet.

Half Your Staff May Be Feeding Client Data to Free AI. You Wouldn't Know Yet.

Magna5 warns shadow AI is shadow IT with higher stakes—49% of workers admit using unapproved tools with contracts and financial data. Owner-led firms need visibility before the leak becomes a client call.

Page 10 of 16. Not sure where to start? Take the free AI assessment.