Cohere launches North Automations, a workflow-orchestration layer for its enterprise AI platform

Cohere has introduced North Automations, a new orchestration layer for its North enterprise AI platform that lets employees describe a multi-step business process in plain language and have several AI agents coordinate to carry it out end to end.

What's new

Cohere describes North Automations as "Cohere's latest step in enabling enterprises to close the AI ROI gap with ease, control, and governance built in from the ground up." The pitch centers on three capabilities: turning complex workflows into simple outputs, staying in control at each step, and governing AI safely at scale. Concretely, that means employees can describe a goal in plain language, the system selects an appropriate model for each step of the workflow, and teams can set approval checkpoints where a human signs off before the process continues.

The feature ships directly into the existing product rather than as a new standalone purchase: "Automations is available today to all North customers," embedded as an orchestration layer inside the platform. Where North's existing agents have generally handled isolated, task-specific requests, Automations is built to chain several agents together across an entire business process.

Context

North launched as Cohere's secure enterprise AI workspace, aimed squarely at organizations that need agents and automation but won't send their data to a shared cloud model — the platform can run inside a customer's own VPC, on-prem, or through Cohere's Model Vault inference layer, with a zero-trust security posture layered in from the start. Cohere has been building out North's surrounding ecosystem quickly, including North Mini Code, its first developer-focused open model released under Apache 2.0, and integrations like a security agent built with Wiz. Automations extends that same enterprise-security framing to multi-step process automation, an area competitors have targeted with standalone workflow and agent-orchestration products.

Why it matters

The generic pitch for enterprise AI agents has increasingly run into a harder problem: most real business processes aren't single tasks, they're sequences of steps spanning multiple systems, and a single agent handling one step in isolation doesn't close the loop. By building orchestration directly into North rather than shipping it as a separate tool, Cohere is betting that its existing security-first customer base — banks, government, and regulated industries wary of sending data to public model APIs — wants that same orchestration without giving up the on-prem and VPC deployment options that got them onto North in the first place. It's a narrower, more defensible niche than the general-purpose agent platforms from OpenAI, Anthropic, and Google, and one where governance and approval checkpoints, not raw model capability, are the selling point.