Inside WRAG

WRAG / 04

How a one-person advisory firm built an AI operating layer.

Jarvis is not a packaged product or a flawless autopilot. It is the working system WRAG uses every day—and the proving ground for the controls we bring to client work.

01

The administrative load

A solo advisory firm still has email, tasks, content, expenses, security, websites, backups, reporting, and client support. The opportunity was not one giant chatbot. It was a coordinated set of small, governed workflows.

What runs through Jarvis

  1. Email and trust controlsThree mailboxes, deterministic filtering, tiered sender trust, disclosure rules, and more than 1,000 logged messages.
  2. Tasks and the daily operating rhythmMicrosoft To Do synchronization, a morning briefing, receipt and expense handling, and recurring search reporting.
  3. System operationsModel failover, uptime and certificate checks, security scanning, resource monitoring, and three-place backups.
  4. Operational memoryCorrections are written down before the next answer, then consolidated into durable operating rules overnight.

The honest part

What the hard lessons changed

Jarvis has caught coordinated phishing behavior and diagnosed client-system failures. It has also caused its own outage while attempting a repair. That incident became a permanent rule: configuration changes require documented recovery paths, backups, and verification.

Useful autonomy is not “the AI can do anything.” It is knowing exactly what the AI may do, what it must record, and when it must stop.

Where it stands

The system runs daily and keeps evolving. Credential failures, failed jobs, and configuration mistakes are treated as operating evidence: fix the immediate issue, then turn the lesson into a rule that prevents repetition.

Your first useful workflow

Want an operating layer built around your business?

Talk with Chuck →