The Business AI Briefing

C.H. Robinson says 30-plus agents cut quote turnaround from roughly 20 minutes to 32 seconds. The map came first.

GPT-5.6 turns connected apps into finished work. The useful question is which tasks need Sol, Terra, Luna—or a human.

A putative class action over AI-generated solicitation calls is a reminder: automation does not outsource a law firm’s judgment.

Shadow AI hides in personal accounts, browser add-ons and copied files. The bill arrives as data exposure, malware and unmanaged dependence.

OpenAI's ChatGPT Work agent gathers context across apps and returns finished sheets, slides, and docs — the shadow-AI upgrade your associates do not need permission to try.

Zoom's standalone AI receptionist bolts onto any phone system from $29.99/month — no migration, no IT ticket, and exactly the kind of agent an office manager can turn on before leadership notices.

Meta's Muse Spark 1.1 API prices at roughly a quarter of top OpenAI and Anthropic tiers — another signal that sticker price is not a model strategy.

KPMG's Q2 AI Pulse Survey finds nearly a third of leaders blind to rising token costs — and half have scaled back when the meter beat the value.

Axios reports the Trump administration approved wide public access to OpenAI's GPT-5.6 models: Sol, Terra, and Luna. For owners, plan continuity first.

Anthropic's intro pricing expires August 31, and a new tokenizer can bill up to 35% more tokens for the same text. Finout breaks down why 'cost-neutral' has an expiration date.

Medicaid, SNAP, and unemployment workflows are being automated — advocates warn that without testing and monitoring, a model mistake strips real people of essential benefits.

MIT NANDA found 95% of integrated AI pilots show zero P&L impact — not because models failed, but because firms pilot without learning, measurement, or workflow fit. The 5% that pay share a pattern SMEs can copy.