Redesigning how hotels manage guest experiences at scale

Solution

Here are some of the key flows

One inbox for guests and staff

The messaging hub unifies guest and staff communication: multi-channel threading, read receipts, templates, and department routing, so nothing falls back to WhatsApp.

Automation that reads like an SOP

Nexus AI presents trigger, condition, and action logic as a structured rule table that matches how managers already write procedures, with a card view on top for review. On the guest side, the same engine powers an AI concierge inside each hotel's own brand.

Content that hotels manage themselves

The CMS lets each property edit its own pages with live preview, so a hotel's digital presence stays on-brand without a designer in the loop.

Glanceable in-room experiences

Smart-room tablets and dining flows were rebuilt around status at a glance and fewer taps to act, for both guests and the staff fulfilling requests.

Overview

A hospitality platform that needed to think at scale.

Intelity builds GEMS 2.0, a large-scale operations platform used by hotels worldwide. Staff across departments use it daily to handle reservations, messaging, workflows, content, and guest requests. The product had accumulated complexity over time, and the design needed a complete rethink to handle that complexity gracefully.

I joined through Tangerine Creative Agency as the lead product designer, responsible for multiple core modules from research and wireframes through to high-fidelity UI and developer handoff.

Problem

Staff were working around the software, not with it.

Hotel operations don't pause. Staff across housekeeping, front desk, F&B, and concierge need to act fast, often under direct guest pressure. The existing GEMS interface mixed too many concepts per screen, lacked visual hierarchy, and had no coherent component logic.

The result: staff were defaulting to workarounds. WhatsApp instead of the messaging module. Paper checklists instead of workflow automation. That's not a feature gap. It's a trust problem.

Initial findings

I audited all eight modules before touching a single screen.

My first step was a full audit of the existing platform. I walked every flow the way a front-desk agent or housekeeping manager would, under time pressure, and logged what broke down: concepts that changed names between modules, actions buried behind inconsistent menus, tables that ordered information by database logic instead of operational urgency.

I also ran an issues-mapping session with engineering, product, and design, collecting every pain point onto one board: prioritization gaps, design misalignment, backlog debt, and resourcing. Grouping the notes by source made the pattern obvious: the three teams were describing the same underlying problem from different angles.

Three patterns kept repeating across modules. Each screen mixed too many jobs. Components behaved differently in different places. And status, the single thing staff check most, was never presented the same way twice. That board became the priority list for the whole redesign.

Goals

Every design goal paired a user need with a business outcome.

Design process

System first, then module by module.

I started with the design system: tokens, input states, table patterns, status chips, and modal flows. Getting the shared language right first meant every module redesign after it got faster and more consistent. Then I worked through the modules in order of operational impact, from messaging and workflows to reservations, content, departments, and integrations.

To keep eight modules moving without chaos, I planned the work as quarterly themes, each focus area running a design phase into a build phase with explicit buffers, and every block linked to its Figma file and Asana/Jira tickets. That roadmap kept design ahead of engineering instead of blocking it.

For each module the loop was the same: wireframe against the audit findings, review with Intelity's product team, iterate, then document the handoff with states and edge cases spelled out. Messaging shows the journey well, from the first structured inbox concepts to the unified hub in the solution above.

Iteration 1

The inbox as a plain table: statuses exist, but conversations still live one click away.

Iteration 2

The conversation panel moves in-context, so staff reply without losing the queue.

Smart Dining shows the approach best: same data, restructured around what staff need first.

Before

Every concept on one screen, no hierarchy, status hidden inside rows.

After

Orders grouped by state, glanceable status chips, and one clear next action.

Outcome

Staff came back to the platform.

After the redesign rolled out, daily active usage rose 34%, and workflow automation recovered around 46 FTE hours per week that staff had been spending on manual coordination. Eight modules now ship from a single design system, which also cut design and build time for everything that came after.

Reflection

My key takeaways and learnings

How I'd use AI on this today