Empowering Property Managers to Self-Train AI in Days, Not Months
VerbaFlo connects UK students with accommodation providers via AI conversations across WhatsApp, email, chat, and voice. But onboarding each new property manager meant 3–6 months of manual AI training by our internal team — delaying ROI, burning 200+ engineer hours per client, and losing 40% of student inquiries to faster competitors.
Lead Product Designer
Product Strategy, UX Flows, Prototypes, Visual Design, Usability Testing
PM, Designers, AI Engineer, Devs, QA
1 Month
Challenge
Opaque AI behavior: PMs couldn’t see why the AI gave certain answers.
No rapid editing: Changes required our team; no sandbox environment.
Uncontrolled deployment: Updates went live without approval—eroding trust.
Scalability limitations: Only one client could be onboarded at a time.
When our AI gave a wrong answer, I had no way to see why—and by the time engineers fixed it, we’d already lost bookings.
Tobias
Property Manger, Sheffield
95%
Pre-Launch Accuracy
67%
Reduction in AI support ticket
200+
Engineer Hours saved per client
Our Solution
Knowledge Base
Property Mangers upload and manage data sources
Real-time processing visibility
Testing Sandbox
Query AI liveReal-time processing visibility
Trace answers to specific sources
Edit inline, see instant changes
Approve changes before going live
Deploy changes
Trace answers to specific sources
Edit inline, see instant changes
