AI News · The field notes
The Daily AI Drop - Evening Edition, November 7, 2026
A massive new model you can run locally, a bold twelve-month prediction, and what actually makes money with AI agents.
A massive new model you can run locally, a bold twelve-month prediction, and what actually makes money with AI agents.
Listen: Listen to this episode
Thinking Machine's Inkling 2 Small is Here
Channel: Fahd Mirza — Watch source video ↗
Thinking Machines released Inkling 2 Small, a mixture-of-experts model with 594 billion total parameters, and Fahd installs and tests it locally on his own hardware. It matters because a model of this scale being runnable by an individual — not just behind an API — changes who gets to experiment with frontier-level AI, and it continues the year's pattern of the gap between open models you can run yourself and closed models behind a paywall shrinking fast.
12 months to the singularity
Channel: Matthew Berman — Watch source video ↗
Matthew makes the case that the next twelve months could be the stretch where AI stops being a tool that helps us work faster and starts transforming how work happens altogether, walking through the pace of recent progress from reasoning models to agents running multi-step projects on their own. It matters less as a prophecy than as a planning exercise: if even half of that progress shows up in the next year, the people who thought ahead about skills and systems that still make sense when AI gets dramatically more capable will have a real advantage.
How to Make Money With AI Agents (What Actually Works)
Channel: TheNextNewThingAI — Watch source video ↗
Andrew Warner studies real businesses already built around AI agents and separates the ideas that are genuinely paying off from the hype, looking at how people use agents to find business ideas, generate leads, and build services clients will pay for. It matters because the pattern behind the winners is refreshingly practical: find a painful, repetitive job a client hates doing, build an agent that does it reliably, and keep a human in charge of the judgment calls — the same shift toward agents doing real, repeatable work that is reshaping software teams and small businesses alike.