Senior Product Designer · Restaurant Platform
April 2026
Portfolio Presentation

Designing for the operator behind the counter.

Mohammed Hussain  ·  Senior Product Designer

01
Intro
Who I am
Background

Product designer who ships end-to-end.

Fintech, SaaS, and AI. Four companies. One pattern — I tend to be the person who goes from zero: zero design team, zero framework, zero system. And stays until it ships.

Now
Birdeye · Senior Product Designer
Owned Reviews, Social, and AI Agents. Led design system for charts and data viz. Among first designers on AI-powered features. 2022–2025.
Before
Yubi · Founding Designer
Built the design function from zero. Owned B2B fintech products end-to-end. 2019–2022.
Earlier
Zoho CRM · Ipedis · Amazon
Complex B2B workflows at Zoho, inclusive design for enterprise clients at Ipedis, brand and marketing at Amazon. 2014–2018.
Intro
Birdeye · context
Before we start

Birdeye is a B2B SaaS platform that helps multi-location businesses manage reviews, social media, and customer engagement — all in one place.

My scope
3 products
Reviews · Social · AI Agents
How I came in
Early AI
Among first designers on AI-powered workflows
Intro
The user
ops
The user I've been designing for

Non-technical. Time-poor. Running complexity across locations.
This is the operator I've been designing for.

At Birdeye
Dental groups, auto shops, franchises

One person running marketing across 5–50 locations. Non-technical. Time-poor. Lives in spreadsheets and group chats.

At OpenTable
Restaurant groups, managers, hosts

One person running reservations, covers, guest data, ops — often across multiple concepts. Non-technical. Time-poor. Same human.

Intro
What I'll cover
Two case studies, one thread

Both shipped solo at Birdeye. Both about making complex operator workflows feel fast.

Case Study 01 · 20 min

AI Content Hub — helps one marketer create a full week of social content across all their locations in under two minutes.

Conversational briefs Multi-format generation Approval workflows
Case Study 02 · 20 min

AI Agent Framework — I designed the container before the feature. Every AI agent Birdeye ships now uses it.

Framework-first Goals · Triggers · Tasks Org-wide adoption

Each case study includes a live Figma prototype walkthrough — I'll share the link when we get there.

Case Study 01
20 minutes
9x
Case Study 01 · Birdeye

AI Content Hub.
One brief. Ten locations. Nine content pieces.

How do you compress a week of multi-location content work into under two minutes — without losing brand voice, approvals, or audit trail?

content ops · approvals · scale
01
Case Study 01 · Content Hub
The problem
The problem

One person running content for five to twenty locations. Every week.

Blank canvas paralysis

Form fields asking "what do you want to post?" didn't help. They needed a thinking partner, not a template.

Brand voice inconsistency

Ten locations, ten slightly different voices. Corporate chasing compliance, franchisees improvising.

Approvals over text and email

Screenshots forwarded over messages, no audit trail. Nobody knew who approved what.

Solo designer · PM & I ran 8 customer calls · 2 shadowing sessions · competitive review

The insight

A form to fight blank-canvas paralysis
just recreates the blank canvas.

Every competitor used form-based briefs. Jasper, Hootsuite, Buffer — all the same. We'd just moved the problem one step earlier.

The fix wasn't a better form. It was not a form.

Case Study 01 · Content Hub
Product walkthrough
Design walkthrough

From conversational brief to published content — in under two minutes.

Covers the full flow: brief → multi-format generation → review & edit → approval → publish & audit.

Open Figma Prototype
Case Study 01 · Content Hub
Impact & decisions
What shipped · what I learned
Time to content
~2 min
from brief to 9 published-ready pieces
Output per brief
9 pieces
across 6 formats, 10+ locations
Approvals
Sig. lift
cycle time vs text/email baseline
Key decision

Shipped chat AND form — not just one

Let users self-select. 65% chat, 35% form, similar completion rates. Both needed.

Key decision

Configurable approvals, not a simple flag

A 10-person team and a 500-location franchise need different routing. One checkbox serves neither.

What I got wrong

Skipped audit trail in V1

Enterprise customers asked week one. Shipped within the first month after launch. For a multi-location team product, audit trail is table stakes.

What surprised me

"Assigned to me" was the most-used screen

We designed for team workflows. People needed their own view first. Teams are made of individuals.

Case Study 02
20 minutes
Case Study 02 · Birdeye

AI Agent Framework.
I designed the container first.

Birdeye had no design language for AI agents. Lead Gen was the first — but I designed the framework first. Multiple agents now use it across the org.

framework-first · reusable primitives · trust
02
Case Study 02 · Agent Framework
The problem
The problem

Leads lost after hours. Handoffs with zero context.

"We'd get chat transcripts at 8am — people asking at 11pm the night before. By the time we called back, they'd gone elsewhere." — Customer interview
The design problem

Every competitor gave operators a prompt box. Wrong mental model for a non-technical operator running ten locations. They don't write prompts — they hire people.

Solo designer · PM · ML Engineer · CS team · customer testing

The insight

Operators don't write prompts.
They onboard people.

The mental model shift: configuring an AI agent should feel like briefing a new hire — not programming a bot.

Goal
What to achieve
Trigger
When to run
Tasks
Steps to take

That's where Goals → Triggers → Tasks came from.

Case Study 02 · Agent Framework
Product walkthrough
Design walkthrough

From framework thesis to 10+ agents running across the org.

Covers: Goals → Triggers → Tasks configuration, reusable tool primitives, Lead Gen agent in action, outcomes dashboard.

Open Figma Prototype
Case Study 02 · Agent Framework
Impact & decisions
What shipped · what I learned
Deflection
~70%
conversations handled without human
Leads captured
~3×
vs pre-agent baseline
Design cost
~60%
reduction per new agent vs scratch
Key decision

Goals/Triggers/Tasks vs a prompt box

Non-technical operators don't write prompts. They configure outcomes. Same model now runs every agent across the org.

Key decision

Showed the activity trace — didn't hide it

Operators need to understand what the agent is doing before they'll deploy it. Transparency = trust.

What I'd do differently

ML as design partner from day one

I treated ML as a cost constraint early. They see model failure modes I'd never catch in usability testing. Start there next time.

Hardest part

Naming things for non-technical users

"Goal" beat "objective." "Task" beat "action." I spent as much time on language as on screens.

Close
Close
warmth
Why OpenTable

At Birdeye I got to design the framework for emerging AI surfaces.
At OpenTable I want to do that for the restaurant operator.

Growth surfaces

Content Hub × restaurants

Reservation demand is still demand generation. Events, seasonal menus, promos, and local discoverability all need tools that help operators move fast without losing control.

Operational systems

Agent framework × ops

The same systems thinking applies to structured workflows like waitlist management, service recovery, table turns, or handoffs between automation and staff.

Hospitality layer

Where it gets exciting

What makes restaurants special is that the system still has to support warmth: remembering preferences, preserving context, and helping staff feel more personal, not more robotic.

The reason this role makes sense to me is that it sits at the intersection of operations, growth, and guest experience. That is exactly where I like to design.

Thank you
Q&A · 15 min
reserve
Thank you

Let's have a conversation.

Happy to go deeper on anything — research method, specific design decisions, trade-offs with ML, post-launch learnings, or how I'd approach OpenTable problems.

hussain.design  ·  hussain.ux@outlook.com  ·  linkedin/hussainhida

End
01 / 32