Product

See the mission, not another chat transcript.

Assign consequential work, step away, and return to durable mission state. The dashboard shows every machine, work item, budget, gate, artifact, and receipt. The assistant you prefer can become the conversational command surface without becoming the authority on whether the outcome succeeded.

System of record

Inspect the real mission state.

See the plan, claimed work, evidence, spend, checks, gates, and terminal outcome without reconstructing events from an assistant conversation.

Bring your assistant

Choose how you ask and steer.

Grex is moving toward assistant-agnostic control over MCP. OpenClaw, Hermes Agent, Grok Bot, Claude, or another capable assistant can become the interface while Grex remains the accountable mission layer.

See pairings and compatibility →

Screenshots below are the live dashboard, lightly redacted.

Nidi overview

The Nidi view lists every machine you own that runs Grex — Manipuli (skill packs) installed, Actor counts, and cost and failures over the last 24 hours, per machine.

Use it to see which Nidus is healthy, which has a rising failure count, and where to add capacity before a mission stalls.

Grex dashboard Nidi view showing two active Nidi with Manipuli, Actor, failure, and cost counts
The Nidi view: machines you own, actor counts, cost and failures per machine.

Missions

The missions list shows every mission across your fleet, its disposition — active, completed, withdrawn — and its work-item count. Each mission links through to its receipt export: what was done, what it cost, and the outcome assertion.

Filter by Nidus to find a specific mission or the cluster around one. Open a receipt directly instead of scrolling a transcript.

Grex dashboard Missions list showing mission names, Nidus, status badges, work item counts, and created dates
Missions with dispositions — active, completed, withdrawn — each linking to its receipt export.

Approval queue

The approval queue surfaces decisions that mission policy reserves for the operator, ranked so the ones that matter most surface first.

Inspect the artifact, evidence, action, and cost, then approve or reject it without leaving the dashboard. The gate remains durable while other independent work continues.

Grex dashboard Decision Inbox showing pending quality gates with accept/reject controls and a phone hand-off link
Policy-gated decisions, ranked and waiting on an operator.

AI Budget

The AI Budget view sets spend caps per Nidus — daily, by default — and the action to take when a cap is exceeded: alert, downgrade to a cheaper model, or pause Actors.

Model Pricing underneath is the per-million-token rate table used to price runs that don't report their own cost, editable in place — no deploy required.

Grex dashboard AI Budget view showing per-Nidus daily spend caps with pause action and a model pricing table
AI Budget: per-Nidus daily caps, pause-on-exceed, and the model pricing table underneath.

Give the computers you control a mission, then return to evidence instead of another transcript.

Share your interest