From Agents to Augmentation: AI’s Next Act

[etfs]

AI agents struggled to live up to expectations in 2025, largely because they couldn’t connect easily to real systems. That’s changing with standards like Model Context Protocol, backed by players such as Anthropic, OpenAI, and Microsoft, which allow agents to interact with tools, data, and APIs.

Still, the dominant theme for 2026 isn’t automation—it’s augmentation. As Kian Katanforoosh notes, AI hasn’t become autonomous enough to replace humans, but it’s becoming powerful at supporting them. The result is likely more hiring, new roles in governance and safety, and AI systems designed to keep people “above the API,” not replaced by it.

[mendez]

agents not working wasnt an ai problem, it was a “cant touch real systems” problem, so mcp actually feels like a big deal