Hello, I’m Bryan. I’m a lead UX designer at Eptura, a SaaS company that makes workplace management software. For over nine years I’ve designed their web apps, mobile apps and room screens. I work remotely from London.

I also design for AI and build with it. I code my own working prototypes and tools.

Next, I’m looking for a lead or senior product design role where I can keep designing, prototyping and shipping.

Selected work

Back to all work

2025 – 2026

Eptura platform & AI : Where do I start?

Every app had its own front door. Then the roadmap said “integrated AI”, and nobody had said how you’d reach it.

Team
Bryan Hayward Lead UX designer Matt Fox VP Product Management Annalisa Recio VP UX
Contribution
StrategyUX designPrototypingUser testing [CHECK]
Platforms
Web
Status
Sign-in:  Shipped AI assistant:  Proposal Adopted as the reference for the version now being built.

In short

Eptura’s platform is four apps, Asset, Engage, Workplace and Visitor, built by different companies before Eptura bought them. Customers subscribe to the mix that suits them, and every one of them has to sign in.

Nobody owned the parts between the apps: how you get in, how you move from one app to another, and where the AI assistant would live. I proposed designs for all three, without being asked. The sign-in and switching shipped. The assistant’s design became the reference for the version Eptura is building now.

How it was: A front door you had to remember

New users received an email with a long, obscure web address and were expected to remember it. Lose the email and you’d lost your way in. The address told the system which customer you belonged to, but the system already knew that from your email and password. It just wasn’t being asked.

Sessions ended after a day, so anyone who booked a desk each morning signed in each morning too.

Once inside, moving to another app meant going back to a home page and choosing a tile. Each app felt like a separate building rather than a room in the same one.

The ask: Integrated AI, undefined

Then the roadmap added “integrated AI”. At that point it was a name, not a working product. Nobody had decided how you’d open it, where it would sit, what it could see, or how it would behave as you moved between apps.

It was the same problem as sign-in, one layer up. Both sat between the apps, so neither belonged to any one team.

What I did: One way in, one place to ask

A sign-in button on eptura.com
Obvious in hindsight. The way in is now the company’s own website, which nobody needs to be told. It also means every returning user passes through the marketing site, which matters to sales: people can’t buy what they don’t know exists.

Email and password, sorted out behind the scenes
Single sign-on where the customer uses it, an email and password where they don’t. The system works out which customer you belong to and what you can see.

Land somewhere useful
One app and you go straight into it. More than one and you go to the app you chose as your starting point. First-time users choose it during a short onboarding, along with their usual office, their team and a light or dark theme. Asking the question once also tells people the setting exists, which is more than a help article buried in settings ever managed.

Switching moved into the app
A dropdown above the navigation lists the apps you subscribe to. Choose another and the menu and page change, but the frame around them stays put. If you only have one app, the dropdown doesn’t appear.

The same frame now runs in Archibus and Serraview, Eptura’s separate products for government customers. The pattern held beyond the four apps it was designed for.

Stay signed in for a week
A checkbox on the sign-in page lets people stay signed in for a week. A daily sign-in treats every user as a risk. The real risk is a device someone else can reach, and that’s a device problem, not a session-length problem. This part was designed but not shipped.

The assistant sits inside the app
The groundwork came from Envision, an earlier Eptura product where executives asked questions of their organisation’s data. I took that idea out of a separate product and put it into the apps people already use.

An assistant in its own window pulls you away from the job you came to do. This one opens in a drawer beside the current page, from a button in the same place in every app. Full screen gives the same conversation more room for longer work.

One assistant, whichever app you’re in
Move from Engage to Asset and the same assistant comes with you, working with that app’s data and your permissions. You meet one assistant, just as you meet one sign-in.

Type or talk
A microphone in the input lets people speak a request rather than type it. Voice is an option, not a requirement. Not everyone wants to talk to their computer in an open-plan office.

Triage, for admins
The assistant also flags problems without being asked: a desk booked but never used, a fault inferred from a pattern of reports. A count on the closed drawer and a short banner lead to a list ranked by priority. On mobile, each one arrives as a notification that needs a single decision.

The buttons follow the type of action. Resolve offers Yes or No. Escalate offers Raise ticket or Dismiss. Information only offers Acknowledge. Every item also has Open chat, for answers that aren’t yes or no. I designed it to line up with the automation work on Asset and Serraview, so one pattern could serve all three.

The same button on the website
An Ask Eptura button sits in the website’s navigation, in the same place as it does in the apps. Signed out, it answers questions about the products. Signed in, it works with your own data. [CHECK: concept only — not live on the site]

What happened: Shipped, then adopted

[TEST block – see 00-testing-framework.md]

Method
Moderated sessions with [n] customers

Task
[TEST]

Task success
[TEST]

Time on task
[TEST]

Ease rating
[TEST]

What people said
“That makes it much easier now. One less thing to remember.” — [CHECK attribution: customer name or role only]

The sign-in and app switching shipped, without the week-long session. [FIGURE: help desk tickets about lost sign-in addresses, before and after — CHECK whether tracked]

I presented the assistant internally as a proposal, and for a while nothing happened. The version now being built uses the same drawer, full-screen mode and triage list, and its design file references my work.

In hindsight: The work between the apps

Nobody asked for any of this, because nobody owned it. The problems sat between the apps, so every team could reasonably say they belonged to someone else.

The flows matter less than the result. Someone signing in for the first time should meet one company, not six, and the second app they open, and the assistant inside it, should feel like part of the same product as the first. That isn’t finished. It is closer than it was.