An IT platform that never forgets a ticket, a server, or a warranty
An internal IT operations platform combining three things that usually live in three separate tools: a helpdesk with real SLA enforcement, uptime monitoring for every internal platform, and a hardware and software asset registry with warranty and licence expiry alerting.
The problem
IT requests arrived by chat, by shoulder tap, and by email, and vanished the same way. Nobody knew if the file server was slow or down. Warranties expired silently until a dead laptop turned out to be three weeks out of coverage. The organisation needed one place where a ticket cannot be forgotten, a degraded service cannot go unnoticed, and an expiring warranty announces itself in advance.
The approach
- Helpdesk with teeth. Every ticket type and priority maps to an SLA matrix. Tickets auto-assign by rules, count down visibly, and breach loudly. Waiting-on-user pauses the clock.
- Uptime monitoring built in. Registered platforms are health-checked on their own intervals and marked up, degraded or down, with response times and 30-day uptime history.
- Asset registry with a memory. Hardware and software with assignments, locations, and warranty or licence expiry alerts at 30, 14, 7 and 1 days out.
- A worker that cannot double-send. A separate always-on service runs the health checks, SLA sweeps and daily expiry alerts. Every alert writes to a log first and dedupes on its identity, so a restart never re-sends notifications.
Built with Next.js server components, PostgreSQL with Drizzle, and Microsoft Entra sign-in, with alerts delivered by email and Teams. Three roles keep it simple: staff raise and follow tickets, IT operations work them, admins own the rules.
The outcome
Requests stopped vanishing, because chat and shoulder taps got replaced by a queue with countdown timers that someone owns. Outages became something IT tells the team about, not the other way around. And procurement got ahead of hardware failures because the platform starts nagging a month before a warranty dies.
What I learned
The alert log matters more than the alert. The first version sent notifications directly, and a worker restart could double-send or lose them. Writing the alert to a log first, then delivering from the log with deduplication, made the whole system boring and trustworthy, which is exactly what an IT platform should be.