The same questions — across three messengers.
A license-plate manufacturer was manually answering repetitive questions about price, availability and delivery across Instagram, Viber and Telegram — with German-speaking clients in the mix.
We launched a multilingual sales agent across the three messengers, integrated with the CRM. It detects the intent behind a message — price, availability, delivery — runs the dialogue and brings it to an order request, replying in Ukrainian or German depending on the language the client writes in. All three channels are synchronized into one stream in the CRM, so history and requests live in one place.
Atypical questions go straight to a manager together with the dialogue context, so nobody re-reads the thread from scratch. The result: one agent runs all three channels, German-speaking clients no longer wait for a German-speaking manager, and about 70% of routine requests close autonomously instead of eating the working day.
What was in the way
- ✗The same questions — price, availability, delivery — across three messengers
- ✗Some clients write in German
- ✗Managers replied by hand in every channel separately
What the system does
Intent detection
Price, availability, delivery — the agent runs the dialogue and brings it to an order request.
Two languages
Replies in Ukrainian or German — whichever the client writes in.
One stream in the CRM
Three messengers synchronized with the CRM: history and requests in one place.
Human handoff
Atypical questions go straight to a manager together with the dialogue context.
How it fits together
How we roll it out.
Requests and access analysis
We study historical requests, verify CRM and API access, and score the automation potential.
A slice of real traffic
The agent runs on part of the live requests with weekly syncs — real dialogues, not a demo.
Metrics and scale-up
We evaluate volume, response speed and escalation share — then expand the agent to full traffic.