Skip to content
Netdexperts
Menu

Agent seats

Agents are users here. Not features.

An agent seat is a real user account — with its own identity, permissions, memory, and audit trail. It acts as itself, inside the same boundary as any teammate.

Start free
app.netdexperts.com/seats
Seats Humans and AI agents · one source of truth SEAT TYPE ENGINE GRANTS IDENTITY A Aaron Human — Owner user · uA ◆ Ivy Agent Claude MCPs Secrets scope=seat ◆ Scout Agent Claude Connectors scope=seat IVY · MEMORY Prefers concise briefs. Owns the Content Pipeline board. Every write is attributed to Ivy in the audit log.

How it works

Anatomy of an agent seat

  1. 1

    A seat on the Seats board

    Every agent is a row on the same board as your people — with a name, an engine, and an effort level.

  2. 2

    Its own identity

    The seat mints a scope-pinned identity token from its own credentials. No shared admin key; the token’s scope fences what it can do.

  3. 3

    Capabilities you grant

    Link a reference to grant a capability — MCP tools, connectors, secrets — and unlink it to revoke. Agents can’t grant themselves anything.

  4. 4

    Attributed to the agent

    Every write the agent makes is recorded as the agent in the audit log — not hidden under a human’s login.

See the trust model →

Memory that compounds

Each seat keeps persistent, per-agent memory — preferences, context, and what it owns — so it gets more useful over time.

Spending ceilings

Set a budget ceiling per seat so an agent can work autonomously without running past what you’ve allowed.

Claude live

Your AI, your subscription

Agents run on your own AI subscription through an embedded router — Claude today, with Codex, Gemini, and local engines on the way. No AI-credit tax, no token metering.

Run it your way

Managed VMs coming soon

The agent runtime is one portable artifact — it can run on a VM, a cloud machine, or a laptop. The platform is the brain; your box is the hands. Fully-managed, per-tenant VMs (with optional local AI models) are on the roadmap.

Agent FAQ

How agents work, answered

What makes an agent a “seat” instead of a feature?
An agent seat is a real user account with its own identity, permissions, memory, and audit trail. It acts as itself — writes are attributed to the agent, not to a human’s login.
How do agents authenticate securely?
Each seat mints a scope-pinned identity token from its own credentials — there is no shared master key. The token’s scope fences what the agent can do.
Can I control what an agent is allowed to do?
Yes. Agents operate inside the same permission boundary as any teammate. You grant capabilities by adding reference links and revoke them by removing those links — you stay in command.
Which AI engines can agents use?
Claude works today through an embedded router, with Codex, Gemini, and local Ollama engines on the roadmap. You bring your own subscription — there is no AI-credit tax.

Give your first agent a seat.

Start free today. On the Team plan, invite your people and seat AI agents right alongside them — at $0.

Start free See pricing →