AgentKit and the Vertical We've Been Anticipating (agent-ci.com)

At Dev Day, AgentKit debuted as a no-code, visual builder for AI agents — sparking debate about whether it threatens traditional developer frameworks. The better takeaway: AgentKit is creating a new vertical, akin to what Zapier did for integrations. It targets people who deeply understand their own workflows (analysts, ops managers, small-business owners) and lets them build automations quickly using a visual interface and built-in eval tools. It’s not trying to replace open-source stacks like LangChain, CrewAI, or PydanticAI, nor the rigorous practices required for production-grade systems; instead it democratizes personal and small-business automation by lowering the barrier to translate domain knowledge into working agents. For the AI/ML community the implications are clear: expect more distributed innovation and many more niche automations as non-developers harness agentic workflows, while open-source frameworks remain essential for reliability, testing, monitoring, and long-term control. Production systems will still demand comprehensive automated evaluations (accuracy, performance, safety, consistency), logging, CI, and the ability to fork or self-host for business continuity. AgentKit’s arrival signals a complementary ecosystem — rapid, user-driven automation at one end and robust, testable, maintainable engineering at the other — with hybrid workflows and integration points likely to follow.