godaddy • product strategy • ai • discovery
Conversations Future Vision
I set out to design an AI business partner. My own research told me to fix the inbox first.
TL;DR Summary
After we shipped an AI inbox assistant, the roadmap moved on — but I believed AI could do far more for small businesses. No roadmap slot and no mandate. I had to make the case myself.
I surveyed SMB owners, ran concept tests, and designed a future vision for an AI business partner built into Conversations. Research redirected the vision. The biggest win wasn't the AI concept — it was surfacing inbox problems leadership hadn't prioritized.
My Role
Lead Product Designer
Year
2026
Partners
Conversations Mobile Designer, AiroHQ Designers, Design Foundations Team
Drag to compare
Context
We shipped an AI assistant. Then the roadmap moved on.
Two GoDaddy products matter here.
Conversations is a unified inbox for small businesses. It brings email, website chat, contact forms, Facebook, Instagram, and Google messages into one place.
AiroHQ is GoDaddy’s AI workspace. Business owners use specialized AI agents for everything from validating an idea to building a marketing plan.
As part of a company-wide AI push, we brought Conversations into AiroHQ with an agent that summarized messages, prioritized conversations, and helped train the chatbot. It was a solid MVP. Then the roadmap moved on.
The Opportunity
I thought users wanted more AI in the inbox. I was wrong.
The more I watched how small businesses worked, the clearer it got: they don’t have an inbox problem. They have a business-operations problem disguised as an inbox. They answer the same questions again and again, miss important requests, and spend their time reacting instead of growing.
What if AI could help owners run their business, not just manage their messages?
What if AI could help owners run their business, not just manage their messages?
Research
Help me. Don’t replace me.
Before designing anything, I surveyed 48 SMB owners on three questions:
- Which inbox tasks do they actually want AI to help with?
- Where do they expect to stay in control?
- What would make them trust an AI assistant?
Most owners wanted AI to sort and prioritize their inbox
Share of 48 survey respondents who mentioned each need
Answers could mention more than one need.
Sorting and prioritizing came up more than twice as often as anything else. One owner put it simply: “Most important e-mails on top, and rank them from most to least important. Kill spam.”
Three themes came back:
- Eliminate repetitive work.
- Organize information, don’t just display it.
- Assist the owner, don’t replace them.
Vision
An AI business partner that knows its place.
With the Conversations mobile designer, I explored multiple concepts and landed on a direction that balanced automation with control. AI would:
- Spot trends across conversations
- Group similar customer requests
- Draft replies in the owner’s tone
- Surface urgent issues automatically
- Suggest bulk replies, while the owner decides what gets sent
One concept we ruled out early: full auto-reply. The “assist, don’t replace” theme made that a non-starter.
AI groups related conversations by topic and surfaces the ones that need attention first — so owners spend less time sorting and more time responding.
Validation
In which my own research proved me wrong.
I interviewed Conversations customers — from casual users to high-volume businesses — and walked them through the vision.
I expected to validate AI features. I didn’t.
Businesses weren’t struggling because AI was missing. They were struggling because they couldn’t trust the inbox. They told us about:
- Missed customer messages
- Unreliable notifications
- System messages buried among customer conversations
- Confusion about what Conversations actually did
“I didn’t know where my messages were going. I finally found them and saw I missed a lead for a $5,500 job. That was bad.”
Research participant — SMB owner
No AI feature fixes that. An AI that summarizes an inbox you don’t trust is still an inbox you don’t trust.
Impact
Build trust while we build AI.
I had a polished AI vision and research that undercut it. I could have pushed the vision anyway. Instead, I argued that the fundamentals couldn’t wait for AI to be finished.
I turned the findings into four principles for everything that comes next:
- Awareness before efficiency. Owners have to know a message arrived before AI can help them handle it.
- Signal over noise. Separate customer conversations from system messages.
- Assist, don’t automate. AI drafts and suggests; the owner decides.
- Keep owners in control. Nothing is sent or hidden without their say.
AI stayed full speed ahead — it was a company-wide initiative, and this research was never going to stop it. It didn’t need to. Revisiting the fundamentals became a design-led initiative running in tandem with the AI work. The two tracks need each other: an AI that drafts the perfect reply is useless if the owner never sees the message.
Reflection
Leading means changing your mind out loud.
I started this project to prove a point about AI. The most useful thing it did was prove me wrong.
The exciting answer and the right answer aren’t always the same. Sometimes the most valuable design work is finding out what not to build yet.
What I’d do differently: I’d talk to existing customers before designing the vision, not after. Testing the vision surfaced the trust problem, but earlier interviews would have found it sooner and pointed the vision at it from the start.