Cohere’s North 2 Wants to Be the Command Center for Enterprise AI Agents, Whatever Model You Run

coheres north 2 wants to be the command center for enterprise ai agents whatever model you run The typical enterprise AI platform expects customers to lock into its own models. Cohere's North 2 takes the reverse approach: use any model you want, and Cohere will provide the space where your AI agents are managed.

The typical enterprise AI platform expects customers to lock into its own models. Cohere’s North 2 takes the reverse approach: use any model you want, and Cohere will provide the space where your AI agents are managed.

At least, that’s the sales pitch. Cohere has reworked its North enterprise platform into a control center for AI agents, which can carry out multi-step workflows independently and remember context between sessions.

Your model, Cohere’s controls

Being model-agnostic is the main selling point. Organizations can use Cohere’s own models, including Command A+, but they can also bring their own.

That flexibility counts most with the customers Cohere is going after. If a bank or a government agency has already standardized on a particular model, it can adopt North 2 without throwing that work away.

Oversight runs through a tool Cohere calls “North Admin.” With it, administrators can manage token usage, set user quotas and control access rights right down to the level of individual agents. Before taking critical actions, agents must also ask for human approval. In regulated industries, that checkpoint will probably draw more scrutiny than anything else on the feature list.

Agents that share what they know

The agents are coordinated by a revamped orchestration system. They draw on shared knowledge libraries and reusable skills, and they hook into tools including Slack, SharePoint and Jira.

From text prompts in the chat window, North 2 can also produce presentations, dashboards and simple apps. Note the word “simple”: Cohere isn’t claiming North 2 can replace your developers.

On efficiency, Cohere says its models use fewer tokens when running on Nvidia’s Blackwell and Hopper hardware. That claim comes from Cohere alone, with no published benchmark to check it against. Until someone tests it independently, it’s best treated as a vendor figure.

Built for people who won’t use the public cloud

Companies have several deployment options. North 2 can run on-premises, in the cloud or fully air-gapped, meaning completely cut off from outside networks.

Cohere is aiming at government agencies and regulated industries that want their AI kept under their own control. Aleph Alpha serves that same audience, and the overlap isn’t an accident. In April, Cohere announced it was acquiring the Heidelberg-based company, which recently reappeared with Kolibri, a German-English language model.

North is also more than a demo. LG CNS and Bell Cyber already run it in production. If your organization is considering North 2, start your questions with those two deployments. They’ll tell you more than any launch announcement will.