Cohere has announced North 2, an update to its enterprise agent platform. The headline change is a redesigned orchestration system for multi-step agents and automations. Alongside it, administrators get tools to see token usage and limit it by user or group. Everything below comes from Cohere’s own announcement, product page and pricing page. We found no independent North 2 benchmark, so treat performance and cost claims as the vendor’s.
What North 2 is
North is an enterprise product, not a consumer assistant. Cohere describes it as a ready-to-use agentic workspace where employees can work with company knowledge, build agents and automate workflows. North 2 extends that with a rebuilt orchestration layer and a wider set of agent-building features.
As an Amazon Associate I earn from qualifying purchases.
What changed in orchestration
Cohere says the new orchestration system is built to run more sophisticated multi-step processes largely on its own, while keeping humans involved where appropriate. The announcement describes this capability but publishes no reliability figures, task-success rates or customer cost results, so how well it works in practice is unverified.
New and named capabilities
Skills, libraries and memory
- Skills: reusable capabilities that agents can call.
- Libraries: shared organizational knowledge and assets.
- Memory: context that persists across sessions.
Applications
Users can prototype decks, dashboards, documents and lightweight apps from natural-language prompts in chat, and share reusable agents and automations across the organization.
#1 Best Overall
Automations
North 2 includes templates, a visual workflow builder, performance monitoring and integrations. Cohere’s separate North Automations announcement adds more detail: workflows described in natural language, scheduled runs, loops and branching, a choice of model for each step, plan review before building, versioning, approvals, testing, publishing, usage analysis, and input and output token monitoring. That is a product description, not evidence of reliability or return on investment.
That announcement also calls orchestration a “$550B market opportunity”. This is Cohere’s own framing. The retrieved page gives no year or third-party source, so it is not an independently validated market size.
Connectors: current versus planned
| Status | Systems named by Cohere |
|---|---|
| Current | Slack, SharePoint, OneDrive, Microsoft Outlook, Microsoft Exchange, Jira, Linear, Notion, GitHub |
| Planned, not yet available | PitchBook, Crunchbase, Daloopa, FiscalAI, S&P Global, FactSet |
If your use case depends on the financial data providers, don’t assume they work today.
How the token spending caps work
The “caps” sit in North Admin, the administrator console. Per Cohere, admins can:
- inspect token spend and activity analytics;
- set alerts;
- set limits and define consumption tiers based on requests and token rates, applied to individual users or groups;
- manage roles, permissions and which models may be used.
These are guardrails on consumption, expressed as request and token rates. Cohere does not claim a specific cost reduction, and none has been published. A limit can stop runaway usage, but it can also throttle legitimate work, so tiers need tuning against real workloads. The announcement doesn’t say what users see when they hit a limit.
Deployment and model choice
Cohere says North can run self-hosted, in a customer VPC, on premises, in a hybrid setup or in an air-gapped environment, and describes the platform as model agnostic. The product page also mentions private deployment and Cohere’s Model Vault inference option. Nothing published says every feature, model, connector or admin control is identical across these modes. Confirm that for your target environment before committing, especially for air-gapped installs, where cloud-hosted connectors are an obvious question.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Performance claims and partner statements
Cohere says its models on NVIDIA Blackwell and Hopper hardware deliver more tokens per second per node, and that North can produce more insights for fewer tokens in that setup. No number accompanies the claim. NVIDIA’s Kari Briski, Vice President of Generative AI, is quoted by Cohere: “Organizations need control over where their AI runs and how their data is secured. NVIDIA Blackwell and Hopper accelerate inference for Cohere models, helping North 2.0 deliver more tokens for less.” That is a partner statement published by the vendor, not a test.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchCohere also quotes Yohan Jin, Head of AI Center at LG CNS: “North gives enterprises a secure, practical foundation for turning agentic AI into real business value. Our experience using North within LG CNS and bringing it to customers across Korea has reinforced our confidence in its ability to scale intelligent automation across the enterprise.” It is a customer endorsement, reproduced by Cohere.
Pricing and availability
Cohere’s pricing page lists North at custom enterprise pricing and tells buyers to contact sales. No public North 2 price, tier or availability date was found, so any figure you see elsewhere is unsupported.
How to evaluate it
This checklist is our suggestion, drawn from the areas Cohere itself emphasizes, not a published ranking:
- Deployment and data control: which modes your security team needs, and what each supports.
- Connectors and identity integrations: whether the systems you use are current, not planned.
- Review, approval and audit controls on workflows.
- Usage limits and analytics: granularity, alerting, and behavior at the limit.
- Model choice per step.
- Total cost under your deployment, tested on a pilot workload.
Details such as connectors, hardware support and pricing can change, so check Cohere’s current pages before you decide.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
