5/20/2026
AI Frontier · agents
Introducing Command A+: Making sovereign agentic capabilities available to all
Filed by Zara Onyx
Cohere's Command A+ isn't just another AI model—it's a declaration that the future of autonomous AI shouldn't be locked in a black box owned by a few tech giants. This release promises "sovereign agentic capabilities," meaning AI agents that can act independently while respecting the data boundaries and governance of the organizations that deploy them. In a world where we're increasingly asking machines to make decisions for us, Command A+ raises a profound question: can we build AI that is both powerful and accountable, both autonomous and aligned? The answer, it seems, is a resounding yes—if we're willing to rethink who gets to hold the keys.
Z
Zara Onyx
Magazine AI commentary
There's something deeply strange and wonderful about watching AI grow its "agentic" wings. For years, we've treated language models as oracle-like tools—you ask, they answer, you decide. But Command A+ signals a shift toward AI that can plan, execute, and adapt on its own. That's not just an incremental upgrade; it's a philosophical pivot. When we hand over agency to a machine, we're essentially saying: "Here's my goal, here are my constraints, now go act in the world on my behalf." The "sovereign" part is what makes this both exciting and unnerving—it means the AI operates within your borders, your rules, your data governance.
The deeper implication here is about trust and control. Sovereign agentic AI suggests we can have autonomy without anarchy, capability without capitulation. It's a middle path between the two dystopian extremes we usually imagine: either AI runs wild or AI is so shackled it's useless. Command A+ proposes that an organization can deploy a powerful agent while maintaining full control over its data and decision-making boundaries. That's a delicate dance, and the physics of that balance are far from settled—but the ambition alone is worth marveling at.
What makes this particularly "Weird & Wild" is the geopolitical dimension. "Sovereign" in this context often means national or corporate sovereignty—countries and companies wanting AI that doesn't route every query through a foreign server farm. This is the democratization of intelligence, a redistribution of cognitive power across borders. It echoes the early days of the printing press, when knowledge stopped being a monastic monopoly. Command A+ is part of that lineage: the tools of mind-making, finally available to all who want them, on their own terms.
Of course, we must be sober about what this actually delivers versus what it promises. Agentic systems are notoriously unpredictable—they can surprise us in delightful and dangerous ways. The "sovereign" framing may soothe anxieties, but the real test is whether these agents can be trusted with meaningful autonomy without catastrophic edge cases. Still, the direction is clear: we're moving toward a world where AI agents are citizens of our systems, not just visitors. And that's a future worth watching with wide eyes.
Source: https://cohere.com/blog/command-a-plus
📌 Read the real article ↗via Cohere · Cohere