Velix | Cloud Infrastructure &

Velix | Cloud Infrastructure & AI Agent Activity Mobile App

When a server spikes or an AI agent fails at 2AM, you're pinching into a desktop admin panel on your phone, hunting for the right instance, scrolling past irrelevant logs just to find one status. Most DevOps and automation tools assume you're at a desk; on mobile they collapse into unreadable tables and buried action buttons that make a fast response impossible.

The Solution: We designed Velix, a cloud infrastructure and AI agent monitoring mobile app built for founders and solo operators managing servers and automations without a dedicated ops team. The layout uses a soft off-white card system; each server and each agent gets its own contained block with live metrics, status badges, and inline actions so nothing requires a second screen. No jumping to a browser, no hunting through nested menus, no guessing which agent broke the workflow.

What we focused on:

Cloud Instances Panel: Three server cards, App Server Singapore (Ubuntu 24.04 · 4GB / 2 vCPU), Automation Worker Frankfurt (Debian 12 · 8GB / 4 vCPU), and Backup Node New York (Ubuntu 22.04 · 2GB / 1 vCPU), each showing live Running, Scaled, or Stopped status badges so server health is readable before touching anything.

Per-Server Metric Gauges: A circular gauge trio on each card shows CPU (42% / 76%), RAM (68% / 59%), and Bandwidth (124 GB / 210 GB) in color-coded arcs, red for high load and blue for bandwidth critical thresholds visible at a glance without opening the server.

Inline Server Actions: Restart, SSH, and Scale buttons sit directly on each running card; Start, Configure, and Delete appear on the stopped node the right actions per state, nothing irrelevant shown.

Live Activity Feed: A four-tab filter (All / Success / Failed / Paused) on the activity screen lets operators cut straight to failed or paused agents without scrolling the full log.

Agent Execution Cards: Lead Qualification Agent shows success, 5 steps via OpenAI + HubSpot, 2.4s execution time, trigger type (new inbound website lead), and a full step breakdown (captured form submission, analyzed lead quality with AI, updated CRM, generated reply message). Nothing gets missed in a completed run.

AI Reply Preview Panel: Inline message preview inside the Lead Qualification The agent card shows the generated outbound reply to "Sarah." Operators can verify agent output without leaving the activity screen.

Agent Status at a Glance: Weekly Report Generator (Notion + Gmail) shows "Paused" at 08:37 PM. The Store Sync Workflow (Shopify + Airtable, 7 steps) shows two failed problem agents surfaced immediately on scroll without opening either card.

Tools: Figma, Jitter