The Command Center — Your Control Room

When to visit the Command Center

You do not need to live in the Command Center. Your agents work on their own. But here is when you should check in:

• Every morning — glance at Control to see if anything needs your approval
• When you get an alert — check Incidents to understand what happened
• Weekly — browse Runs to see what your agents accomplished
• Occasionally — look at Overview for the big picture

The Command Center overview page showing agent status cards, system health indicator, and recent activity feed

The nine tabs

Overview

The landing page when you open Command Center. It shows:

• System health — a single green/yellow/red indicator for the whole platform
• Agent status cards — each agent with their current state (active, idle, or needs attention)
• Recent activity feed — the last 20 things your agents did, newest first
• Connection status — whether ServiceTitan, QuickBooks, and Gusto are all connected

This is your 10-second health check. If everything is green, you are good.

Agent status cards showing Penny (active), Dash (idle), Sterling (active), and their last action timestamps

Control

This is the most important tab for day-to-day use. Control shows everything that needs your attention:

• Pending approvals — payroll runs, invoice batches, email drafts, or any action waiting for your sign-off
• Active directives — instructions you or Tower have sent to agents
• Queued actions — tasks agents have prepared and are ready to execute once approved

Each item shows what the agent wants to do, why, and the supporting data. You click Approve or Reject for each one.

Control tab showing two pending approvals: a payroll run from Sterling and an invoice batch from Penny

Agents

A detailed view of each agent in your system. Click any agent to see:

• Current status — what they are doing right now
• Last check-in — when they last reported in
• Recent tasks — their last 10 completed actions
• Error rate — how often their tasks succeed vs. fail
• Autonomy settings — what they are allowed to do without your approval

Agent list showing Penny Banks, Dash, Sterling, Monroe, and Chase with status badges and last activity

This tab is useful when you want to understand a specific agent's workload or check if an agent is having trouble.

Heartbeat

The heartbeat is the 6-minute check-in cycle. This tab shows a real-time timeline of every check-in:

• Green dots — agent checked in successfully, everything normal
• Yellow dots — agent checked in but reported a warning
• Red dots — agent missed a check-in or reported an error
• Gray dots — agent is offline (expected if outside business hours)

Heartbeat timeline showing a row of green dots with one yellow dot for Sterling at 2:15 PM

If you see red or yellow, click the dot to see what happened. Most of the time, yellow means a temporary API hiccup that resolved on its own.

Incidents

When an agent encounters something it cannot fix on its own, it files an incident. This tab shows all open and recently resolved incidents:

Each incident card shows what happened, what the agent tried, and the current status. Tower automatically reviews incidents and pushes fixes — most low and medium incidents resolve themselves within a day.

Incidents tab showing one medium-severity incident about a QuickBooks sync delay

Runs

A complete history of every task every agent has executed. Each run shows:

• Agent — who ran it
• Task — what they did (e.g., "Morning Report — Lexington", "Estimate Follow-up")
• Start/end time — when the task ran
• Duration — how long it took
• Status — success, partial, or failed
• Token usage — how many AI tokens were consumed

Use this tab when you want to verify that a specific task ran, see how long things take, or audit what happened on a particular day.

Runs history table showing recent agent tasks with timestamps, durations, and status badges

Skills

A catalog of every skill (capability) your agents have. Each skill shows:

• Name — what it does (e.g., "Invoice Batch Prep", "Estimate Follow-Up", "Payroll Prep")
• Agent — which agent owns it
• Schedule — when it runs
• Autonomy level — whether it requires approval
• Success rate — historical success percentage

This is a reference page — useful when you want to know what your agents are capable of or when you are deciding which tasks to promote to autonomous.

Skills catalog showing a grid of agent capabilities with schedule and autonomy indicators

Training

Agents learn from every task they run. The Training tab shows:

• Recent learnings — things agents discovered in the last 7 days
• Playbook entries — self-healing recipes agents have created
• Accuracy trends — how agent accuracy has improved over time
• Training batches — sets of similar tasks agents use to calibrate their work

You do not need to interact with this tab regularly. It is here for transparency — so you can see that your agents are getting better over time, not just repeating the same patterns.

Training page showing recent learnings and an accuracy trend chart

Admin

System-level configuration and diagnostics. This tab is for advanced users:

• Agent configuration — edit agent roles, schedules, and permissions
• Autonomy settings — promote or demote task autonomy levels
• Connection diagnostics — test ServiceTitan, QuickBooks, and Gusto connections
• Token budgets — set spending limits for each agent and task type
• System logs — raw logs for debugging

Most owners never need this tab. It is here for when you want to fine-tune the system or troubleshoot a persistent issue.

Admin panel showing connection diagnostics and token budget settings

A typical day in the Command Center

Here is what most owners do: