Kono — Conversational Appointment Bot for 1Q Bankers
A Facebook Messenger AI bot that let customers book in-person banking visits in natural language — Korea's first conversational scheduling PoC for a major bank.
KEB Hana Bank's '1Q Banker' program deploys mobile advisors directly to customers — but booking a visit required navigating call centers or legacy web channels with limited capacity.
Kono Labs built a conversational scheduling bot on Facebook Messenger to remove that friction. The bot handled natural-language date input, matched customers to available 1Q Bankers by district, confirmed appointments, and issued calendar invites — 24/7, with no human agent.
What I owned: · Conversational UX design — dialogue flows for booking, rescheduling, and cancellation · UI/UX for card carousel interactions (banker selection, time slot picker) · Natural language interaction design — date/location parsing UX, error handling · User flow documentation and PoC process specification · Collaboration with KEB Hana Bank 1Q Banker Center on pilot scope and rollout
Project Phases: Phase 1 — Flow Design & Spec (Jul–Aug 2017): Core user flows, dialogue scripts, UI component spec. Phase 2 — Beta PoC (Aug–Sep 2017): Pilot with volunteer 1Q Bankers in 5 Seoul/Gyeonggi districts.
Booking a banker visit required calling a call center.
The 1Q Banker Center's existing booking channels — internet, mobile app, and a call center — couldn't scale to meet demand. Call center agents were limited, and digital channels lacked the flexibility for conversational scheduling.
Customers who wanted a banker to visit them faced friction: no clear channel, no real-time availability, no self-serve rescheduling. The bank needed a low-cost, always-on booking layer that felt natural to use on mobile.
Three intents. Natural language. Zero app install.
Chose Facebook Messenger as the channel — already on customers' phones, no app install required, and supporting rich card UI (carousels, quick replies, outlinks).
Designed the bot around three core intents: book, check, and change — surfaced immediately as quick-reply buttons so users never faced a blank input. Layered in natural language handling for dates and locations so the experience felt conversational, not form-like.
Kept the PoC scope tight: 5 districts, volunteer bankers, Google Calendar as the backend — validating the model before any internal system integration.
Book a banker visit in under 2 minutes, in plain Korean.
Designed a full conversational flow across three user journeys:
1 · Book a visit — customer selects district → Kono shows available bankers (carousel with photo, name, coverage area) → customer picks a time slot from NLP-parsed availability → inputs meeting location as free text → provides callback number → receives confirmation + ICS calendar file
2 · Check appointment — returning user detected → bot surfaces existing booking with full details and options to cancel or modify
3 · Change or cancel — customer-initiated or banker-triggered (with automatic push notification to customer if banker changes the schedule)
UI highlights: · Banker selection via image carousel cards with district labels · Time slot quick replies (3 options per screen) · Free-text location input with no format constraint · ICS calendar download + SMS/email confirmation dual delivery
Korea's first conversational banker-visit bot — PoC shipped and validated.
· Delivered Korea's first conversational appointment bot PoC for a major retail bank · Demonstrated viability of SNS-channel banking services without call center dependency · Validated NLP-based scheduling UX (natural date/location input) under real user conditions · Established a replicable service model for Kono Labs' subsequent enterprise chatbot contracts · Provided KEB Hana Bank with interaction data and a working prototype to inform full-channel rollout planning
