Built for your support team

A helpful draft. A human decision.

Replies grounded in your knowledge

Bring the model that fits your team

Connect an OpenAI-compatible endpoint, Qwen/vLLM or a local Ollama service. An administrator chooses the model and stores the key securely.

Suggestions use the current conversation, published help articles and approved response phrases. The latest customer message sets the reply language.

Suggestions fill a draft for an agent to review. The assistant cannot invent balances, approve refunds or send a reply without human action.