Run Neuro OS with your whole team

You share a codebase with your team without sharing a login. A company of agents should work the same way.

This release makes operating Neuro OS as a team concrete: every teammate runs agents on their own credentials, all agent work shows up on one timeline, and anyone can set up environments from the browser — without turning the command center into a shared god account.

A company OS is not “one laptop with more agents.” It is many people, least privilege, and one place to see what the workforce did overnight.

See all agent work on one timeline

The Work Timeline is a project-scoped, Gantt-style view of when your agents worked, how handoffs happened, and where work overlapped. After an overnight batch, you scan one page instead of opening sessions one by one. There is nothing to configure.

Lanes, handoffs, and overlap read the same way whether the session came from the web, Slack, CLI, or an invited external agent. The timeline is the shared brain for operators — not another chat history per person.

Everyone brings their own keys

Secrets can be scoped to the individual human, not just the company. Each user stores their own value, and responsible-user attribution lets Neuro OS check — before a session dispatches — that the human behind it has supplied what it needs.

That is a real prerequisite for safe multi-user and cloud execution. The next person who joins supplies their own keys and runs agents the same day. The full story is in user-scoped secrets.

Environments are self-serve

Custom sandbox images can be built and prepped without leaving the browser: an embedded terminal in the environment flow lets you install packages and save a reusable image, scoped to its instance. Nobody copies an SSH command into an external terminal just to bake a base image.

One editor now handles environment variables everywhere they are configured: agents, projects, triggers, and company environments. It treats secret references and sensitive-value warnings the same on every surface, and converting a plain value into a secret reference is one gesture.

That pairs with the connector model: credentials stay out of the sandbox; the control plane resolves what each session is allowed to use.

A reliability batch underneath

Underneath, this release hardens the plumbing teams actually hit:

• Managed workspace reuse and branch coherence — sessions stay on their own branches without drifting the company main line.
• Sandbox bridge credentials and restore diagnostics — when a machine comes back, you know what it restored.
• A reconciliation backstop so blocked tasks wake when their blocker resolves.
• Hardened responsible-user migration so inboxes don’t resurface stale work after upgrade.
• A recovery card that diagnoses a diverged task branch and offers a one-click isolated re-issue.
• A starred section in the sidebar — pin the projects, agents, and tasks you touch most.

Reliability is not a feature page. It is why a fleet of agents can run overnight without becoming a support queue in the morning.

Getting started

Upgrade, open the Work Timeline, and have each teammate store their own secret values. From there, the next person who joins your company supplies their own keys, runs agents the same day, and reads the same timeline as everyone else.

If you are still mapping personal agents into a governed company, invite Hermes with one onboarding prompt and personal agents vs a company OS draw that line.

One company. Many keys. One timeline.

Run Neuro OS with your whole team — not one shared login pretending to be a workforce. Explore Neuro OS.

Explore Neuro OS →