All Chats
GTM Workspace was adding new customers fast. Busy sellers needed a quick way to find and reference their past AI conversations. My cohort study became the business case that moved a deprioritized feature into development, shipped across three products. Lead designer, with three PMs, two engineers, and a design partner.
Product Context
GTM Workspace is ZoomInfo's unified AI platform for enterprise sales teams, and Chat is how sellers work inside it. In natural language they analyze data and generate strategic content, with the AI drawing on CRM records, ZoomInfo intelligence, engagement history, and the web. Every conversation can also spin off artifacts, the AI-generated documents sellers plan and act from, such as account plans and buyer engagement maps. The product was launching fast, with new customers onboarding daily.
The Problem & Business Opportunity
GTM Workspace was launching fast, and the chat list shipped as an engineering stopgap. One endless list. No filters, no persistence, no way to find and resume a past conversation. Artifacts, later rebranded as AI Docs, had the same problem: a small pill shape gave no visual information to tell one from another.
Visual Teardown: List View of Chats in GTM Workspace
What the Research Revealed
Partnering with a second designer, I ran eight moderated A/B sessions with internal GTM Workspace users, comparing the legacy flat list against a table view with company tags, artifact counts, and message dates. Sellers live in spreadsheets. The research confirmed the table matched how they think. Visual artifact previews removed the old navigation dead-end, so no one had to open a chat just to recall what an artifact was.
Quotes from Research Findings
-
“I'm kind of like a rows and columns guy... When I look at this, it's just a lot more logical... So this is a phenomenal upgrade in my opinion.”
—Account Manager II -
“I feel like you can sift through a lot more chat history and information faster.”
—Account Manager III -
“I think the consolidated view with the date and the table view is much easier for me to process quickly visually... Everything I do is in tables pretty much, right?”
—Account Manager III -
“We saw a few moments ago when you asked me if I had created artifacts. I had to go scroll through and figure out where I'd done it. This shows me here in one place.”
—Account Manager II
My Role & System Execution
One pattern then had to hold across three products (GTM Workspace, GTM Studio, and the legacy SalesOS) and two design systems. The AI chat was the same Angular component in every product. In GTM Workspace it ran as an iframe inside the React shell. Handoff was a full responsive spec across five breakpoints from 960px to 1920px, in both design systems: Muze's classic components for SalesOS and GTM Studio, and MUI's muted blue-gray set for GTM Workspace. The Muze-themed products had no standalone All Chats page, so a canvas container connected to the chat panel carried the feature instead.
Muze versus MUI Theming
Responsive Specs
Static previews shown at this screen size. The interactive All Chats prototype is available on desktop.
Table View
Scan by thread name, company (or a multi-company View), last message date, and artifact count. Columns match how sellers already work: in tables.
Artifact Discovery
Select the artifact chip to preview every document a thread produced, without reopening the chat.
Row Options
Rename and delete are now added options, only one click away in each row.
Company Filters
Filter by company (recently viewed first), view, and last message date for easy scanning.
Active Filter
Applied filters narrow the table to one company.
Search Field
Keyword search stacks on top of any active filter.
Loading State
Skeleton rows hold the layout during load to reduce blank-state anxiety.
Results & Impact
When this research wrapped in Q4 2025, the improvements had full support from our Product Manager and development team. Then priorities shifted: PMs rotated off, the roadmap changed, and the half-built table sat behind a feature flag. It had stalled on personal preference, not research-backed evidence.
I kept the research alive through the transition. When a new PM took over the Chat team, I brought the data back to the table: 7 of 8 Workspace users found the table view a clear upgrade and far easier to scan. The feature was unblocked immediately and fully shipped at the beginning of Q2 2026.
