The infrastructure for AI agents is coming together fast. While everyone debates what agents will eventually do, the companies building the pipes and platforms are shipping the boring stuff that makes agents actually work in production.
01
Google opens free AI agents course to 1.5 million developers
Google relaunched its free 5-day AI Agents course, this time focused on "vibe coding" with agents. The previous version attracted 1.5 million learners. The curriculum covers building agents with Google's CLI, connecting them to external APIs, adding long-term memory, implementing security guardrails, and testing reliability.
Why it matters: When Google teaches 1.5 million developers to build agents the same way, that becomes the de facto standard. Every startup building agent tools will need to work with whatever patterns this course teaches.
Box CEO Aaron Levie explains why AI progress is slowing enterprise adoption
Box CEO Aaron Levie shared a counterintuitive insight on The MAD Podcast: AI breakthroughs are happening so fast that they make obsolete whatever enterprises just implemented, which actually slows down overall deployment. "The problem is the breakthroughs keep happening faster than the customer can implement any kind of standard architecture," he said.
Why it matters: Your IT team just spent six months integrating GPT-4 into workflows, and now GPT-5 makes that setup look outdated. The faster AI improves, the more gun-shy enterprises become about committing to any particular approach.
Vercel CLI goes dependency-free to power "cloud for agents"
Vercel CEO Guillermo Rauch announced that Vercel's CLI is now a self-updating binary with zero external dependencies. He called the CLI "one of the key interfaces enabling the cloud for agents" and said the company is shipping CLI changes more frequently than ever as it gets embedded in more systems.
Why it matters: Agents need to deploy code, not just write it. If Vercel becomes the default deployment platform for AI-generated applications, that's a significant moat in the agent infrastructure stack.
France's Mistral AI launched Voxtral TTS, an open-weights text-to-speech model designed for voice agents. The company describes it as "fast, instantly adaptable" and capable of producing "lifelike speech."
Why it matters: Open voice models have lagged behind text models in quality. If Mistral cracked realistic voice synthesis that anyone can download and customize, that removes a major barrier for developers building voice-first applications.
Swyx met with the ACM President to discuss collaboration between the AI Engineer community and the Association for Computing Machinery. Industry spotlights from the CAIS conference will be presented at the AI Engineer summit next month, with more formal collaboration planned.