Virtual Agent Quality Audit demo
Find where the chatbot failed and fix the knowledge behind it. 1:13
What the demo shows
A system solution for digital channel teams that audits chatbot conversations for failures.
Six agents read every chatbot conversation, mark the turn where the bot failed and replay it: a loop on “I didn’t get that”, a wrong answer about a charge, a hand-off that never arrived. Each failure is traced to the call the customer made next, and each cluster becomes a proposed missing skill built from customers’ own words, or a knowledge fix with the source that proves it. The digital channel owner approves skills; the knowledge manager approves fixes.
Transcript
- Virtual Agent Quality Audit reads every chatbot conversation, and marks where the bot failed.
- The bot dashboard says 76% contained.
- Only 61% were actually helped.
- The rail ranks the failures.
- I didn’t get that loop spoil the most chats.
- Sunday evening, a customer tries to send her water meter reading.
- Lumi misses it twice, then explains how to read the dials.
- On the 3rd, I didn’t get that.
- the replay stops. Missing skill, take a meter reading, 37 chats.
- The proposal lists what the skill must collect in customers' own words.
- Hannah approves, and the bot team gets the request.
- Wrong answers get fixes too.
- Reconnection isn't free under the 2026 tariff.
- Build on Botminds, agents, data, screens, and governance on one platform.
Transcribed from the film’s narration. Captions are on the player.