Nori vs Claude Code: why a team platform if we already use Claude Code?

Quick answer

Nori and Claude Code are not rivals: Nori runs Claude Code. Claude Code on its own is a harness for one engineer at a terminal, plus Anthropic's cloud features for that engineer's account. Nori puts the same Claude Code, on the Claude subscription you already pay for, onto shared cloud machines that the whole team drives from Slack, schedules and webhooks, with one shared skillset and memory. Choose Nori when the work belongs to a team, not a laptop.

What is Claude Code?

Claude Code is Anthropic's coding agent harness. Run locally, it works in your terminal, on your checkout, with your credentials, for as long as your laptop is open. It is very good at that job, and nothing on this page argues against it.

Anthropic also offers Claude Code in the cloud. As of November 2026 its docs describe durable cloud schedules, API triggers and GitHub triggers as a research preview, and customer-run environments as a public beta for Team and Enterprise plans in which Anthropic keeps the control plane. A separate product, Claude Tag, brings Claude into Slack channels as a public beta for Team and Enterprise, funded from an organization balance. That product has its own page: Nori vs Claude Tag.

Published prices on claude.com/pricing: Pro $20 per month, Max $100, Ultra $200, and Team standard seats at $25 per month or $20 per seat billed annually.

What is Nori?

Nori is a cloud platform for coding agents, built for engineering teams. It runs any agent harness, Claude Code included, in isolated cloud sandboxes that the team drives from Slack, Discord, the web or the CLI. Cron schedules, one-time reminders and webhooks start the same kind of session, with run history and approval paths. Org skillsets, shared memories and scoped integrations mean every session starts with how your team ships.

You connect your own Claude subscription, API keys or LLM gateway. Nori never charges for tokens, so the Claude plan you already have is the one that funds the work.

Meet Jiro

Nori also ships Jiro, a ready-to-work AI engineering teammate. Jiro reads your repositories, learns how your team ships, and takes requests in Slack from day one. Teams that want a packaged teammate start with Jiro; teams that want plain Claude Code in the cloud run that. Both use the same machines, memory and automations.

Nori vs Claude Code, head to head

The two prices are not alternatives. A team on Nori still holds Claude plans, because those plans fund the model usage. Nori's plan price covers the machines and the team layer around them.

What changes when Claude Code runs on Nori

Take one small bug. With local Claude Code, a support lead reports it in Slack, an engineer copies the report into a terminal, waits, and pastes a pull request link back. If the engineer is in a meeting, nothing happens.

On Nori the thread is the session:

#support-escalations

Priya: @nori invoices exported after 23:00 UTC show tomorrow's date.
       Repo is billing-api. Fix it and add a test.

Nori:  Reproduced. format_invoice_date() used the server's local time.
       PR: billing-api#412 (1 file changed, 1 test added, suite green).

Marco: @nori reviewer asked for a timezone-aware fixture. Please add it.

Nori:  Pushed a follow-up commit to the same PR.

Three things are different. Priya is not an engineer and did not open a terminal. Marco picked up a session he did not start. And the session read the team's CLAUDE.md, skills and shared memories from the org skillset, not from one person's home directory.

The same session can start with no person at all. A schedule such as 0 6 * * 1-5 can run Triage last night's failed CI runs on main and post a summary to #eng-ci, and a Sentry webhook can open a fix pull request for review. See scheduled coding agents and Sentry webhook to fix PR.

Enterprise deployment: where does the control plane run?

A coding agent has a place where commands execute and a control plane that decides who may start a session, which credentials it gets and what is kept. Moving the first does not move the second. See what an agent control plane is.

Anthropic's customer-run environments are a public beta for Team and Enterprise as of November 2026, and Anthropic retains control in that model.

For Enterprise, Nori deploys the full stack, control plane included, into your own AWS VPC or Azure VNet, or on-prem, with your identity provider (SAML, SCIM, OIDC), your LLM gateway and your compute. Hosted Nori customers get a dedicated, single-tenant control plane and data store per organization.

When to choose Claude Code on its own

• You are one engineer, or a few, and the work happens at the keyboard in pair-style sessions.
• Nobody outside engineering needs to ask the agent for anything.
• You have no recurring work to schedule, or you are comfortable building on features Anthropic labels research preview.
• Every engineer maintaining their own configuration is acceptable, and you only want Anthropic's harness.

Most Nori teams keep local Claude Code for exactly this. The question is where background, scheduled and team-requested work goes.

When to choose Nori

• Requests arrive in Slack from engineers, product managers and support, and you want the pull request to come back in the same thread.
• Work must keep running when laptops sleep: overnight jobs, CI triage, dependency updates, webhook-driven fixes.
• You want one org skillset and one shared memory, so the tenth engineer's sessions behave like the first's.
• You want to run Claude Code today and keep the option of Codex, Gemini CLI or another harness without changing platform.
• You want to keep paying Anthropic directly for inference, with no token charge from the platform.
• Security needs the control plane, not only the runners, inside your own cloud.

How to move a team from local Claude Code to Nori

1. Start the trial. Sign up for the 30-day free trial (no card) and connect GitHub and Slack.
2. Connect your Claude account. Sign in with the Claude subscription you already use, or add an API key or gateway. Usage stays on your Anthropic bill.
3. Move the config that already works. Copy the CLAUDE.md, .claude/skills/ and subagents from your best-configured engineer into the org skillset. Every session now starts from it.
4. Write the setup script once. Put the commands a new laptop needs (for example pnpm install and cp .env.example .env) in the org setup script, so each machine boots ready to run tests.
5. Run one real task from Slack. Pick a small bug and ask in a channel: @nori fix the null region bug in the CSV export and add a test. Review the pull request as usual.
6. Schedule one recurring job. Take the chore someone does by hand each week and give it a cron line, for example 0 9 * * 1 for a Monday dependency update that opens a pull request.
7. Keep local Claude Code. Nothing is uninstalled. Engineers use the terminal for hands-on work and send everything else to Nori.

The longer product walk-through is in run Claude Code in the cloud for your whole team.

Frequently asked questions

Sources

Anthropic pages were read on November 11, 2026. Prices and preview labels change; check the linked pages before you decide.

• Claude Code docs: Routines (cloud schedules, API and GitHub triggers)
• Claude Code docs: Self-hosted environments
• Claude pricing
• Claude Tag docs: Overview
• Claude Tag docs: Memory
• Jiro, Nori's AI engineering teammate