How can we help?

Choose your AI models

Updated

Talvo runs on OpenRouter, so you are never locked to a single AI vendor. You can pin the model behind your copilot and knowledge-base drafting; customer-facing AI answers stay on Talvo's tuned models.

How it works

By default, Talvo routes each task to a model tuned for it. You can pin a specific model in Settings → General → AI model and switch at any time. A pinned model applies to the AI copilot and to knowledge-base drafting only. Customer-facing auto-resolve answers, translation and background analysis stay on Talvo's tuned models.

Shown with our demo workspace.

What it costs

Model usage is included in your Talvo plan. There is no separate AI bill and no API key of your own to manage. You are billed $0.49 per AI resolution beyond your plan's allowance, and conversations your team handles in the shared inbox are not charged that way.

Choosing a model

A bigger model is not automatically a better support agent. What determines answer quality, in order:

  1. Whether a published article covers the question. No model can answer from content that is not there.
  2. How clearly that article is written.
  3. The model.

If the agent is handing off too often, the model setting is not the lever: customer answers don't use your pinned model, and the gap is in the knowledge base. The AI performance view will tell you which is which: a low grounding rate points at your articles.

When switching is worth it

  • Copilot replies or knowledge-base drafts are accurate but stilted, and your brand voice is not landing.
  • You prefer a particular model for the drafts your team reviews and edits.

Switching is safe

A pinned model affects future copilot replies and knowledge-base drafts only. Nothing is retrained, no history is reprocessed, and your knowledge base is untouched. If a change does not help, switch back.

Was this article helpful?