Build a bank support chatbot with clear handoff rules

Commerzbank uses Gemini in its Bene chatbot to answer customer chats and pass tougher cases onward.

Done with Gemini· Intermediate

What they did
Commerzbank was an early user of Google’s Customer Engagement Suite and used it to create Bene, its in-house banking chatbot. The company later added Gemini to improve the chat experience. The available description does not spell out the build process, the data Bene can access, or exactly when a human takes over.
How it went
Bene now deals with more than two million chats and settles 70% of inquiries. The source does not give a starting point for comparison, define what counts as resolved, or provide timing or cost figures.
Tips
Treat vendor claims as a starting point, not a final proof. Before using a setup like this, check how escalation works, what sources the bot can read, and how resolution is measured.

Try it yourself with Gemini

You are helping me design a customer support chatbot for [bank or business]. Using the policies, FAQs, and product rules I paste below, draft: 1) the bot’s role and boundaries, 2) which requests it should handle end-to-end, 3) which requests must be handed to a human with a short handoff summary, and 4) the tone it should use. Do not invent capabilities beyond the materials I provide. Before any action that would send a message, submit a form, delete data, or change an account, stop and ask for my approval. Finish by listing 15 realistic customer questions with the best response for each so I can test it.[paste materials here]

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on cloud.google.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.