Fig. 01 · Every AI suggestion paired with its reasoning - trust through transparency.
Context
Enterprise sales teams sit on top of CRM records, contract histories, pricing rules, and competitor signals, but rarely combine those inputs in real time when preparing for a customer conversation. Krystal AI set out to turn that fragmented institutional memory into live, actionable recommendations, ingesting CRM and contract data to suggest which leads to prioritize, what pricing to position, and which risks to call out before a call.
Goal
- Create an AI-powered tool to provide actionable recommendations to the sales team.
- Help sales professionals create more effective pitches and contracts.
My role
- Was solely responsible for the user experience, from user needs definition to final UI specifications.
- Delivered detailed concepts and interactive prototypes for the development team.
Key decisions
Recommendations with an explicit "why"
In sales environments, opaque AI is easy to ignore because reps must justify their decisions to managers and peers. I designed every recommendation in the interface to carry an explicit explanation of why it was being suggested - the signals it drew from, patterns it matched, and the confidence level attached. This emphasis on transparency aimed to build trust so reps could confidently defend the system's guidance in real conversations.
Built for skimming, not deep analysis
Sales reps typically use tools in short bursts - five minutes between calls, not hour-long analysis sessions. I structured the core UI around compact, scan-friendly cards, each with one clear, primary action, rather than dense dashboards. Deeper views lived on secondary screens for the minority of users who wanted to investigate patterns further, keeping the main experience lightweight and time-respectful.
Fig. 02 · Customer performance view with C-Score tracking and contract metrics.
Impact
- Directly helped the sales team generate the right leads at the right time.
- AI-driven suggestions proved valuable in closing deals more efficiently.
Takeaway
What I took forward
I came away with a strong belief that an AI recommendation is only as valuable as a salesperson's willingness to say it out loud and stand behind it; designing for that moment of verbal defense helps align everything from explanation patterns to interaction density.
Tools & methods
Photoshop
Sketch
Axure RP
Design Thinking Methodology
Invision
Agile
User Interviews
Goals
Persona Creation
Journey Mapping
Task Analysis
Wireframe & Hi-Fis
Prototyping