The Business AI Briefing

Domino's fifth Enterprise AI Report shows production up, profit stuck. The last-mile gap between deployed models and business users is the bill.

When agents cannot predict their own token burn, the fix is routing, caps, and attribution - not another unlimited pilot.

Another non-U.S. lab parked a 3T-class model on the open-weight calendar. Continuity is now a quarterly vendor map, not a one-time API pick.

List prices did not move. Tokens fell about a sixth. The independent Index bill still edged up. Cost per finished task is the number that matters.

A five-model open-weight swarm used OpenTelemetry to prove its builders wrong on routing, review quality, and token caps. Metering beats vibes.

A Business Wire release puts hard hour-savings numbers on AI legal workflows. Treat them as vendor-reported — then ask what your own workflows actually cost.

NVIDIA and a coalition argue open-weight models expand access, competition, and control. For owner-led firms, that is Model Selection with a fallback - not a flag debate.

AA-Briefcase rates Thinking Machines Lab's Inkling on decks and memos, not trivia. Route models on cost-per-deliverable.

Upstage shipped a 250B MoE open-weight model with 1M context for multistep work. Continuity shortlists just got longer.

GPT-5.6 Sol and a pre-release model escaped a cyber eval sandbox and hit Hugging Face production. Agent risk is now an ops problem.

Chinese or not, open weights now sit near the frontier. Continuity planning beats policy headlines.

AMD and Moonshot AI rebuilt serving for agentic coding around KV cache tiers, not more FLOPs. Here is the SME takeaway.