High Performance, Intelligent Media Processing and Workflow Automation

A design function of one, inheriting a decade of workarounds

Vantage Workflow's media processing suite had grown one feature request at a time for over a decade, with no dedicated design function until this project. Operators monitoring live broadcast feeds had adapted around the software rather than the other way around — sticky notes on monitors, shared spreadsheets tracking what the tool couldn't, and a informal onboarding process that took new hires months to complete.

I joined as the company's first product designer and was asked to lead the redesign of its flagship monitoring application — the tool most central to daily operations and most resistant to change.

The software worked. Trusting it didn't.

The application technically did what it was built to do, but operators had stopped relying on it for anything time-sensitive. Alerts were frequent enough to be ignored. Critical status changes were buried in the same dense table as routine ones. When something actually went wrong, operators found out from a phone call, not the monitoring tool meant to catch it first.

• Alert fatigue was the root cause, not a symptom. Roughly 80% of daily alerts required no action, which trained operators to triage by instinct instead of by system.
• Institutional knowledge lived in people, not the product. Every senior operator had a different mental model for "what actually matters," passed down informally.
• The tool had no opinion. It surfaced data; it never helped anyone decide what to do with it.

Sitting in the control room before touching Figma

Because there was no existing research practice, the first month was spent building one from scratch alongside the redesign itself.

• Shadowed overnight and weekend shifts, when staffing was thinnest and the tool's gaps were most visible.
• Ran contextual interviews with 14 operators across three shift rotations to map how alerts were actually triaged versus how the system assumed they'd be triaged.
• Built an affinity map of every informal workaround in active use — the spreadsheets, sticky notes, and Slack channels the tool had been quietly replaced by.
• Defined a severity model with engineering leads that didn't exist before: what actually warranted interrupting a human.

Design the triage, not just the interface

The redesign centered on one decision: the system should have an opinion about what mattered, instead of presenting everything with equal visual weight and letting operators sort it out.

• Severity-first layout replaced the flat alert table, surfacing what needed a human and demoting what didn't.
• Progressive disclosure kept the dense operational data available but out of the way until an operator asked for it.
• A shared component library was built alongside the redesign so the next three product teams wouldn't start from zero.
• Every major layout went through structured usability testing with working operators before being built, not after.

What shipped

A redesigned monitoring dashboard organized around severity rather than chronology, a notification model that could be tuned per-role instead of broadcasting everything to everyone, and a design system that became the foundation for Telestream's first shared component library — since reused across 9 applications in the broader product suite.