It’s Tuesday
Here’s what’s worth your time today.
Anthropic has launched a new enterprise agents program that lets companies deploy Claude-powered agents across finance, HR, legal, and engineering. These agents can plug into internal tools like Gmail, DocuSign, and company databases, then execute tasks directly such as generating financial analysis, drafting job descriptions, or preparing technical specs. Companies can customize agents and distribute them internally through private marketplaces, similar to deploying software. The shift is clear. AI labs are no longer selling chat interfaces. They are positioning themselves as the operating layer inside enterprise workflows, putting them on a collision course with traditional SaaS tools.
Supported by Airia
Join Airia’s AI Agent Hackathon ahead of the Australian Race weekend. The Virtual AI Agent Hackathon runs February 23–March 1, giving you seven days to design autonomous AI agents that elevate the fan experience. Top projects will be highlighted ahead of the first race weekend of 2026.
Anthropic claims Chinese AI firms including DeepSeek, Moonshot AI, and MiniMax used millions of Claude conversations to replicate its capabilities through model distillation. The companies allegedly created over 24,000 fake accounts and ran roughly 16 million prompts focused on coding, reasoning, and agent workflows. Distillation itself is common, but Anthropic says this was done at industrial scale to bypass the cost and time of training frontier models. The implication is bigger than one company. Model outputs are becoming strategic assets, and protecting them is emerging as a new front in the global AI race.
Despite the hype, OpenAI COO Brad Lightcap says AI hasn’t yet penetrated core enterprise workflows. Individuals use ChatGPT heavily, but integrating AI into real business processes spanning teams, systems, and approvals remains difficult. OpenAI’s new Frontier platform is designed to help companies build agents tied to actual business outcomes, not just per-seat licenses. Partnerships with consulting giants like McKinsey, Accenture, and BCG signal how much hands-on work adoption still requires. The takeaway is simple. AI is everywhere at the individual level, but the true enterprise transformation is only beginning.
Meta has signed a multiyear deal to buy up to $100 billion worth of AMD GPUs and CPUs, enough to power massive new AI data centers. The agreement includes equity incentives tied to performance milestones, showing how strategic these partnerships have become. Meta is racing to build infrastructure for what Mark Zuckerberg calls “personal superintelligence,” while also reducing reliance on Nvidia’s expensive chips. The bigger shift is structural. Control over compute is becoming one of the defining advantages in AI, and hyperscalers are securing supply years in advance.