Cohere launches North 2 AI agent platform with spending controls

Cohere today released North 2, a rebuilt platform for deploying AI agents inside large organizations. The update introduces a redesigned orchestration engine and new administrative controls that cap token spending, addressing what the company describes as a trade-off enterprises have accepted until now.

SiliconAngle reports that the platform overhaul targets multistep agent workflows, suggesting Cohere is positioning North 2 as a bridge between experimental agent projects and scalable, cost-managed deployments. The spending caps appear designed to prevent runaway token consumption—a concern that has surfaced in recent coverage of rogue agent risks and containment strategies.

The timing aligns with a broader shift in the AI agent space. When StartupReader covered Nvidia’s open-source safety platform and ServiceNow’s "control tower" pitch last week, both companies framed agent governance as a question of risk mitigation. Cohere’s approach reframes the same challenge as a budgeting problem, implying that enterprise buyers are now moving from proof-of-concept to production and demanding predictability.

North 2 also arrives as competitors expand their own agent strategies. AMD’s reported local AI platform and Exabeam’s "agentic security operations" updates suggest a market where hardware, software, and governance layers are converging around AI agents. Cohere’s focus on orchestration and spending controls may signal where enterprise adoption is stalling: not on capability, but on cost and control.

Sources: siliconangle.com