Tech Souls, Connected.

Why Cohere Thinks North Is the AI Agent Enterprises Can Finally Trust

A new private-deployment platform aims to make AI safe for sensitive industries

The data security roadblock for AI agents

AI agents promise to reduce repetitive tasks and streamline workflows. Yet adoption in enterprises, regulated sectors, and government agencies has been slow.
The sticking point? Data security. Many organizations fear that sensitive trade secrets or customer information could be exposed — or worse, end up in a model’s training data.

Introducing North: Cohere’s privacy-first AI agent platform

Canadian AI firm Cohere believes it has an answer. Its new platform, North, is designed for private deployment so data never leaves the customer’s control.
Unlike typical deployments on Azure or AWS, North can run:

  • On-premise servers
  • Hybrid clouds
  • Virtual private clouds (VPCs)
  • Even air-gapped environments

Co-founder Nick Frosst emphasized that North can run on minimal infrastructure — “even on a GPU in a closet” — requiring as few as two GPUs.

Security built in from the ground up

North comes with granular access control, agent autonomy policies, and continuous red-teaming to detect vulnerabilities.
It meets GDPR, SOC-2, and ISO 27001 compliance standards, with regular third-party security testing.

Beyond privacy: versatile AI capabilities

North includes ready-to-use chat and search functions, enabling:

  • Customer support automation
  • Meeting transcript summarization
  • Marketing content generation
  • Market research and data retrieval from internal and web sources

Every response includes citations and a reasoning chain so employees can audit and verify outputs.
The platform’s core technology is powered by a variant of Cohere’s Command model, trained for enterprise reasoning, and supported by Compass, its multimodal search stack.

Integrations and automation

North can connect with workplace staples like Gmail, Slack, Salesforce, Outlook, and Linear, plus any Model Context Protocol (MCP)-compatible system for industry-specific tasks.
As users grow comfortable, Frosst notes, there’s a “smooth transition” from augmentation (supporting human work) to automation (handling tasks independently).

Market momentum

Cohere has raised $970 million to date, most recently at a $5.5 billion valuation.
Pilots are already underway with RBC, Dell, LG, Ensemble Health Partners, and Palantir.
Its acquisition of Ottogrid earlier this year adds specialized market research automation capabilities.

Share this article
Shareable URL
Prev Post

Two Charged in Alleged AI Chip Smuggling as Nvidia Pushes Back on Kill Switches

Next Post

Tinder Bets on AI and College Dating to Revive Growth

Read next