RCI Agent Hub // AI Hub
A command center
for your AI workforce.
Agent Hub is the control plane for a fleet of AI coding agents working across many projects at once. See what every worker is doing, put a human in the loop on anything that matters, and keep spend, access, and audit under control as the fleet grows.
Built & operated by Roderick Consulting
- svc01 / aura-ingestRUNNING
Reconciling drilling feed, batch 0417
84k ctx$1.42 - vdi-03 / oxbo-reportNEEDS APPROVAL
Awaiting approval — schema migration
31k ctx$0.38 - svc02 / themis-nlRUNNING
Building visualization for cost report
52k ctx$0.91 - worker-07 / legajos-qaIDLE
Idle — heartbeat OK
4k ctx$0.04
AI agents got powerful fast. Most teams can't see what theirs are doing.
One AI coding agent on one project is easy to watch over someone's shoulder. A fleet of them, running across a dozen projects, is not. Without a control plane, teams lose visibility into what agents are actually touching, lose control over which operations require a human, and lose cost discipline as usage scales past what anyone is tracking by hand. The agents don't get less capable. The organization just stops being able to answer basic questions about them.
Agent Hub is the control plane, not another agent.
Agent Hub doesn't write code and doesn't review it. It sits above the fleet — Claude Code and other AI coding agents working across many projects — and gives the humans running that fleet one place to see it, govern it, and account for it: live status per worker, a governance layer that gates the operations that matter, a standing budget on spend, and an audit trail that survives the conversation.
Worker sessions, exactly as an operator sees them.
Every worker in the fleet reports its status, current task, model, context usage, and running cost in real time. This is an illustrative capture of that view — a small fleet working across a handful of projects.
| Worker | Status | Task | Cost |
|---|---|---|---|
Worker-01 API Builder | WORKING | Adding pagination to /v2/orders endpoint web-api-refactor | $2.14 |
Worker-02 Test Runner | WORKING | Running regression suite, batch 4 of 7 billing-service-migration | $0.87 |
Worker-03 Code Reviewer | BLOCKED | Awaiting approval — schema migration on prod DB billing-service-migration | $1.53 |
Worker-04 Docs Writer | WORKING | Drafting migration guide for v3 API consumers docs-pipeline | $0.22 |
Worker-05 Infra Auditor | IDLE | Idle — heartbeat OK, awaiting next scheduled run infra-cost-audit | $0.04 |
Worker-06 QA Agent | WORKING | Reproducing onboarding drop-off on Android 14 mobile-onboarding | $1.01 |
- 0:02
Worker-01 Opened PR #482 on web-api-refactor
- 0:09
Worker-03 Requested approval: schema migration
- 0:14
Worker-06 Filed reproduction steps + screen capture
- 0:26
Worker-02 Regression batch 3 of 7 passed (211/211)
- 0:41
Worker-04 Draft migration guide saved to docs-pipeline
Illustrative capture — worker names and tasks are examples, not a live feed.
Everything a fleet operator actually needs.
A real-time activity feed of every AI worker across every project: status, current task, context and token usage, and running cost, all in one view.
Governance levels — auto, notify, soft-block, hard-block — gate sensitive operations. One-click magic-link approvals over email or WhatsApp keep a human in control without living in a dashboard.
Per-project AI spend, budgets, and alerts, so a fleet that scales in headcount doesn't scale into a surprise invoice.
Chain specialized agents — discovery, build, test, review — with structured hand-offs between stages instead of one agent doing everything alone.
Teams and granular role-based access mean people see only the projects they're granted. Every action lands in a full audit log, with usage analytics on top.
Heartbeat monitoring and a stable per-worker identity mean a restarted or rescheduled agent is still the same accountable worker, not a new unknown.
Register, monitor, gate, audit — the same loop for every agent.
Every worker in the fleet moves through the same control loop, whether it's running a discovery pass or pushing to production. The loop is what makes a fleet of agents governable instead of just fast.
- 01Agents register
Each worker checks in with a stable identity and a heartbeat, wherever it runs.
- 02Work is monitored live
Task, context usage, and cost stream into the fleet view in real time.
- 03HUMAN GATESensitive ops are gated
Governance levels route risky actions to a human before they execute.
- 04Everything is audited & costed
Every action, approval, and dollar lands in an immutable, per-project ledger.
Control that scales with the fleet, not against it.
Speed and control aren't a trade-off here. Agent Hub is built so a fleet can grow in size and autonomy while the humans responsible for it keep exactly the amount of oversight they need — no more manual babysitting, no less accountability.
Sensitive operations don't execute silently. Governance levels route them to a person, and a magic link is enough to approve or deny from a phone.
Every task, approval, denial, and dollar is written once to an append-only log. What happened on a project can always be reconstructed, not reconstructed from memory.
Access is scoped per project and per role. A worker or a teammate sees exactly the projects they're granted, and nothing else, by construction.
Budgets and alerts are set per project up front, so spend is a dial you set, not a number you discover on an invoice.
We use Agent Hub
to run our own fleet.
Agent Hub isn't a concept slide. It's the control plane Roderick Consulting runs internally to coordinate the AI coding agents building and operating the rest of our product line — Aura, Themis, Oxbo, Data Bastion, and the other systems in production today. The same governance, visibility, and cost discipline you'd be evaluating for your own team is the discipline we depend on for ours. That's the showcase: infrastructure serious enough that we bet our own delivery on it.
See your fleet, governed from one screen.
Request a demo and we'll walk you through live fleet monitoring, approvals, and cost tracking on a real set of agents.