Feature Design + Prototype
2 weeks
Apr 2026
Employee Feedback Software
Designing Live Q&A for Suggestion Ox: A New Feature That Had to Feel Native
Suggestion Ox
Prototype
Fully interactive
Every state built and clickable, not mocked up flat
Tooling
State switcher
Jump to any screen state to evaluate and plan them all
Handoff
Style guide for dev
Backend team builds the feature from a defined system
The Challenge
Suggestion Ox had product-market fit. Their anonymous suggestion boxes gave employees a safe way to speak up, and companies relied on them. That's a strong position, and a good reason to expand carefully.
The opportunity was a new feature for their existing users: Live Q&A during meetings. Employees submit questions in real time, the crowd upvotes the ones that matter, and moderators run the session live, with AI helping to moderate and answer. It's a natural extension of what Suggestion Ox already does best: surfacing what people actually want to say.
The hard part wasn't the idea. It was everything the idea implied.
- It had to feel native. A new feature that looks or behaves like a bolt-on erodes trust in a product people already rely on. Live Q&A had to feel like it had always been part of Suggestion Ox.
- Live means many states. Moderator and attendee views, questions coming in, being approved, upvoted, answered, dismissed, sessions opening and closing. A live feature has a lot of states, and each one needs to be designed, not discovered in production.
- AI had to help, not take over. Moderators needed to stay in control while AI did the heavy lifting of triage and drafting answers.
You can't spec your way to confidence on something this interactive. You have to feel it.
The Approach: A Fully Interactive Prototype
Rather than design the screens flat and hope they added up to a coherent live experience, we built the whole thing as a fully interactive prototype. Real questions flowing in, real upvoting, real moderator actions, real AI moderation, all clickable.
That let everyone judge the feature the way users eventually would: by using it. Does approving a question feel fast enough? Does the moderator view stay calm when questions flood in? Does the AI help or get in the way? Those questions only have honest answers when the prototype actually behaves.

The Live Q&A hub: every session in one place, live, paused, or ended, with moderation status at a glance.
The Live Q&A hub: every session in one place, live, paused, or ended, with moderation status at a glance.

The live host control centre: AI-flagged questions to review, upvotes, starring, and session controls in one view.
The live host control centre: AI-flagged questions to review, upvotes, starring, and session controls in one view.

The participant view: submit a question and upvote, so the questions the room cares about rise to the top.
The participant view: submit a question and upvote, so the questions the room cares about rise to the top.

Session setup, prototyped in full: moderation modes (including AI moderation) and identity options from anonymous to SSO.
Session setup, prototyped in full: moderation modes (including AI moderation) and identity options from anonymous to SSO.
A Live Moderator Control Centre
The heart of the feature is the moderator experience. Running a live session is a real-time job, and the dashboard had to make that feel manageable rather than frantic.
We designed it as a control centre:
- See everything at a glance. Incoming questions, what's been approved, what's answered, and what's rising through upvotes, all in one live view.
- AI as a co-pilot, not autopilot. AI flags sensitive or suspicious questions for review and summarizes the session afterward, but the moderator decides what goes live. Control stays with the human running the room.
- Upvoting as a signal. The audience surfaces the questions that matter most, so moderators spend attention where the room wants it.
Designing Every State: The Prototype Switcher
A live feature is really a collection of states. An empty session before it starts. A quiet trickle of questions. A flood. A question approved, answered, dismissed. A session closing. Miss one and it shows up as an awkward moment in front of a live audience.
So we built a prototype state switcher: a way to jump instantly to any screen state in the flow. Instead of clicking through a whole session to reach the "questions closed" view, you could jump straight to it, evaluate it, and refine it.

Every state was designed, including the quiet ones: a session before it starts, waiting for its first question.
Every state was designed, including the quiet ones: a session before it starts, waiting for its first question.

Getting a room in: a shareable participant link and QR code, prototyped as a real modal over the live dashboard.
Getting a room in: a shareable participant link and QR code, prototyped as a real modal over the live dashboard.

The end state: an AI-generated summary, key themes, and action items once the session wraps.
The end state: an AI-generated summary, key themes, and action items once the session wraps.
This made planning the feature genuinely tractable. Every state could be seen, compared, and signed off deliberately, rather than hoping the edge cases would work out once it was built. It turned "did we handle that case?" into something you could just go look at.
Making It Feel Native, and Ready for Dev
Feeling native isn't luck. It comes from careful reuse of the existing system's patterns, spacing, and behavior, so Live Q&A reads as Suggestion Ox rather than a feature grafted on from somewhere else.
Because the prototype was fully built out, and paired with a style guide, the backend team got more than a design to interpret. They got a working reference for how every state should look and behave, and a component system to build it from. The interactive prototype became the specification.
What Made This Work
Serving strength, not chasing novelty. Live Q&A extends what Suggestion Ox is already loved for. The best new features deepen product-market fit rather than distract from it.
Interactive from the start. A live, real-time feature can't be judged from static frames. Building it as a working prototype meant decisions were made on how it actually felt to run a session.
A tool for the job: the state switcher. Complex features have complex state. Building a switcher to jump between every state turned an overwhelming design problem into a reviewable checklist.
Built to hand off. The prototype plus style guide gave the backend team a resolved, buildable target, so development started from certainty instead of guesswork.
“A prototype-first approach was great. It's nice to be able to iterate fast, and it would be nice to have a way for the LLM to translate that back into the actual app seamlessly.”

Andy Berkowitz
Founder & CEO of Suggestion OxSuggestion Ox Live Q&A is a clear example of prototype-first feature design paired with an AI-ready design system: build the whole experience as a working prototype, design every state deliberately, make it feel native, and hand engineering a resolved, buildable direction.
Adding a significant feature to a product people already rely on? Schedule a discovery call and let's prototype it before you build it.