Streamlining the Enterprise Discovery Engine
Designing for Opportunity
How might AI remove the friction that's quietly slowing down enterprise workflows — and in doing so, help a growing SaaS company stop losing customers?
The Solution
A proactive AI filtering system integrated directly into MS Teams, shifting the approval process from days to minutes.
The Challenge
Enterprise renewals dropped by 5% because the platform felt "clunky" and disconnected from where users actually work.
My Role:
UX / Product Designer
01 BUSINESS CONTEXT
A Company at an Inflection Point.
Connex.io — a customer insights platform — was setting its sights on enterprise clients. The pitch was compelling: deeper data, smarter reporting, AI-powered decisions. But the numbers in early 2023 told a different story.
Customer renewals were slipping. New business targets weren't being met. And internally, there was unresolved tension about whether investing in AI capabilities was worth the risk. The product needed to earn its place in larger, more demanding organizations — and fast.
02 THE CUSTOMER
Zenith Tech Finds Connex.io a Bit Clunky.
Zenith Tech is a global tech company using Connex.io to surface customer insights for product and marketing decisions. Their teams are capable and data-literate — but the platform is slowing them down rather than accelerating them.
Two users sit at the center of the problem. Their individual frustrations add up to a systemic one.
Zenith Tech users find Connex.io repetitive— especially pulling data from other tools and waiting on feedback. AI isn't the goal; AI is the fix. By automating the tedious parts, integrating data sources, and accelerating approvals, Connex.io can rebuild trust with its users and give the business a path back to growth.
Mapping where business goals break down.
I explored how Connex.io's business objectives and the day-to-day needs of Zenith Tech's users were connected — and where they diverged. Through problem mapping, I traced the exact moments where the workflow created friction, delay, and frustration.
The pattern was consistent: users were spending time on tasks the platform should be handling automatically — pulling data from external tools, reformatting results, and waiting on approvals that had no real reason to be slow.
03 — Comprehending the Workflow
Where Alex's Day Goes Wrong.
Mapping Alex's journey through the platform surfaced specific pain points — the moments where the tool created work instead of removing it.
NOTEABLE INSIGHTS
→ Alex must build reports with insights for his manager — but the tools don't support this efficiently end-to-end.
→ Finding relevant information requires searching across topics with no consistent naming structure or labels to guide him..
→ His process routinely starts in Connex.io, then spills into other platforms because Zenith Tech data isn't all in one place.
POTENTIAL AI OPPORTUNITIES
→ Use AI to scan selected documents and surface or suggest relevant insights automatically.
→ Utilize AI to pull reports containing relevant information, reducing manual search time.
→ Integrate reports from other platforms into one unified view inside Connex.io.
→ Enable users to open and access document files from external wiki sources without leaving the platform.
03B — FRICTION IN THE FLOW
Days Lost. Days Reclaimed.
Before proposing any solution, I mapped the current workflow in terms of time — not features. The question wasn't "what does the platform do?" but "how long does it take, and where does time disappear?"
The answer was stark. An 8-day end-to-end cycle could become 5 — not through heroic effort, but by removing the waiting.
04 VISUALIZING THE TIMELINE
From Mild to Wild: Scoping AI's Role.
Rather than jumping to a single solution, I mapped a range of AI integrations — from conservative enhancements to more ambitious reimaginings. This "mild to wild" framing helped surface the right level of intervention without over-engineering or under-delivering
Mild
MIDDLE
Mild
The mild-to-wild exercise confirmed AI's highest-value role: report generation. Data researchers spend hours gathering and formatting — AI can own that work, freeing managers to focus on decisions rather than documents.
05 AI WITHIN THE DISCOVERY PHASE
Search across document sources without leaving the platform. A side panel surfaces details and AI-generated insights inline — no tab-switching, no context loss. Lowers friction with minimal behavioral change required.
Conversational search with follow-up responses. Results allow Alex to dig deeper through natural language queries — the search becomes a dialogue rather than a lookup. Insights can be bookmarked for later report compilation.
A unified AI search across all connected databases. One query, everything. Generated insights can be saved to a "Report Clipboard" that compiles automatically into a shareable, reviewable document — eliminating manual report assembly entirely.
06 SHIFTING THE FOCUS
The Approval Problem is Really a Visibility Problem.
After solving for the discovery phase, I turned to the other half of the delay: Trisha's approval workflow. She isn't hard to reach — she's just getting the wrong signal through the wrong channel.
Email notifications require constant inbox monitoring, interrupt context, and create a mental overhead that compounds across a manager's day. The friction isn't in the work itself — it's in the handoff.
07 – 08 THE SOLUTION
Meet People Where They Already Work.
The answer wasn't a new tool — it was a smarter handoff using a tool Zenith Tech already relies on: Microsoft Teams. By routing Connex.io notifications through Teams and surfacing reports directly in a web browser view, the approval loop collapses from an 8-day process to one that fits in the 12 minutes between meetings.
The same work. Significantly less waiting.
The redesigned workflow doesn't ask users to change how they think or work — it removes the parts that were never their job to begin with. AI handles the repetitive; Teams handles the handoff; Connex.io holds the result.
Newly Enhance Workflow
Discovery Phase: 1 - 2 days
1. Connex.io Search Results, Alex clicks “AI-filter”
3. Alex views insight of interest, Clicks “Save Insight”
Approval Phase: 1 day
Tricia finished a meeting and has 12 minutes before her next meeting. She decides to check MS teams for any project updates.
5. She receives a MS Teams notification on her laptop. It is a Connex.io notification to approve Alex’s work.
7. Clicks the comments button to leave notes on selected items; clicks “Send Feedback” when they are done
8. Alex receives Teams notification from Trisha regarding feedback received on the report
6. Teams notification opens web browser view of Alex’s report.
Approval Phase: 1 day
Alex receives a MS Teams notification, informing him that Tricia provided feedback. They click the notification and Tricia’s comments are available for Alice to review. Now, they are able to move onto the Analyze Phase.
8. Redirected back into Connex.io; Comment icons appear on items that have feedback.
1–2
Days for the Discovery Phase
(down from 5+)
The real insight here wasn't about AI. It was about recognizing that slow approvals and lost customers often share a root cause: people working in parallel when they should be working together. Design's job was to close that gap.
09 NEWLY ENHANCED WORKFLOW
2. AI-filter creates new tab with a more refined set of results
4. Insights saved compile in “Saved Sources” tab which will serve as a report page for Tisha.
1
Day for the Approval Phase
(down from 2–3)
0
New tools required —
MS Teams already in use