Skip to main content
Cohere North 2 AI Control Room interface, compatible with various models, displayed on a monitor.

Editorial illustration for Cohere’s North 2 AI Control Room Works With Any Model

Cohere's North 2 AI Control Room Works With Any Model

• 4 min read

Cohere rebuilt its enterprise platform around a single idea: let AI agents run the work, not just answer questions. The company's new release, North 2, turns its chat interface into something closer to a dispatcher for automated tasks, coordinating agents that remember context across sessions, pull from shared knowledge libraries, and reuse defined skills instead of starting from scratch each time. It plugs into Slack, SharePoint, and Jira, and can spit out presentations, dashboards, and basic apps from a text prompt without leaving the chat window.

The platform runs wherever a company needs it, on-premises, in the cloud, or fully air-gapped, and it doesn't lock customers into Cohere's own models. Command A+ is the default, but organizations can swap in their own systems. A new "North Admin" layer gives IT staff control over token budgets, user quotas, and permissions down to the level of individual agents, with human sign-off required before any agent takes a consequential action.

LG CNS and Bell Cyber are already running North in production environments. The pitch is aimed squarely at government agencies and regulated industries that want AI systems they can fully control, the same market Cohere's recent acquisition target, Aleph Alpha, has spent years courting.

North 2 is model-agnostic. Alongside Cohere's own models like Command A+, organizations can plug in their own. Admins use "North Admin" to manage token usage, user quotas, and access rights down to individual agents.

Why this matters

North 2's bet is that enterprises don't want another chatbot, they want a dashboard that can actually run their Jira tickets and Slack threads without someone babysitting the output. The model-agnostic pitch is the smart move here: Cohere knows Command A+ alone won't win the budget conversation against OpenAI or Anthropic, so letting IT swap in whatever model a compliance team already trusts lowers the switching cost to basically zero. The approval gates and token-level admin controls matter more than the flashy agent orchestration, honestly, because that's the stuff that gets a platform past a CISO rather than killed in procurement.

For founders building on top of Cohere, this is a signal that the real competition isn't model quality anymore, it's who controls the plumbing, the permissions, and the audit trail. Running on Blackwell and Hopper hardware is table stakes at this point, not a differentiator. Watch whether North Admin's granularity actually survives contact with a 5,000-person org chart, because that's where most "enterprise control rooms" quietly fall apart.

Common Questions Answered

How does Cohere's North 2 differ from a traditional AI chatbot?

North 2 transforms a chat interface into a dispatcher for automated tasks rather than just answering questions. It coordinates AI agents that remember context across sessions, pull from shared knowledge libraries, and reuse defined skills, enabling it to run actual work like managing Jira tickets and Slack threads without constant human oversight.

What does it mean that North 2 is model-agnostic?

North 2 can work with any AI model, not just Cohere's own Command A+ model. Organizations can plug in their own preferred models, which lowers switching costs and allows enterprises to use models that their compliance teams already trust.

Which platforms and tools does North 2 integrate with?

North 2 integrates with Slack, SharePoint, and Jira, allowing it to coordinate work across these enterprise tools. It can generate outputs including presentations and dashboards, making it a comprehensive control room for enterprise automation.

What administrative controls does North Admin provide?

North Admin allows administrators to manage token usage, user quotas, and access rights down to the individual agent level. This granular control enables enterprises to maintain security and cost oversight across their AI agent deployments.

Why is North 2's model-agnostic approach strategically important for Cohere?

By allowing enterprises to use any model rather than only Command A+, Cohere removes a major barrier to adoption against competitors like OpenAI and Anthropic. This strategy acknowledges that Cohere's own models alone may not win enterprise budget decisions, but providing flexibility on model choice significantly reduces the switching cost to zero.

LIVE09:34Cohere’s North 2 AI Control Room Works With Any Model