Claude Sonnet 4.6 raises the floor, Open Claw's founder joins OpenAI, and the rise of the 'vibe coder'

Anthropic releases a cheaper, smarter Sonnet. Open Claw gets absorbed into OpenAI. And new survey data shows half of AI coders are non-engineers.

A new sweet-spot Claude model, the Open Claw drama gets a satisfying resolution, and the data is in: AI coding is going mainstream beyond engineers.

Anthropic launches Claude Sonnet 4.6 at Opus-level intelligence

My take: Same intelligence as Opus but $3/$15 per million tokens, with a 1M context window. Best-in-class for agentic financial analysis and office tasks. This is the model your business should be defaulting to for everyday automation: it's now cheap enough to use everywhere. If your team is on Claude Pro or Opus by default, look at switching.

OpenAI hires the creator of Open Claw

My take: Peter Steinberger, the developer behind the viral personal-agent project, has joined OpenAI to lead personal agents. The Open Claw project moves to an open-source foundation. Big strategic win for OpenAI, which had been lagging in the agent narrative. Watch for personal AI agents to get a major push from OpenAI in the next two quarters.

Anthropic vs. The Pentagon tensions escalate

My take: Anthropic's terms of service prohibit military and surveillance use, and the Pentagon is threatening to blacklist them. This is the first major test of whether AI companies can hold ethical lines against government pressure. It will shape what kinds of contracts are even possible for the major labs. If you sell to government, watch closely.

Apple ramping up AI wearables

My take: Smart glasses, a wearable pin, camera-equipped AirPods, all under development as hands-free interfaces for a more capable Siri. The bet is that voice and ambient AI become the next major UI shift. Probably 18-24 months out for serious consumer impact, but worth designing your customer touchpoints with this in mind.

Half of people using AI for coding are not engineers

My take: Survey data shows 49.5% of those using AI for code work outside traditional engineering and IT roles. Marketing, ops, finance, HR. The cost of building small custom internal tools is falling fast. If your IT team has a 6-month backlog of small requests, you should be enabling business users to solve their own problems with AI.