AI agents & answers

Names, models, instructions, languages and reliable answers.

12 step-by-step guides

Change an agent’s internal name

Rename the agent in your team’s list without confusing it with the name visitors see in the messenger.

Choose an available model for the agent’s workload, then compare answer quality, response time and credit usage.

Create instructions that define the assistant’s role, boundaries, response style and next step when it cannot help.

Use small, measurable edits instead of replacing a working instruction set with a long list of exceptions.

Make the first chat interaction clear and give visitors useful ways to begin.

Control whether an agent is available to public widget and developer chat requests.

Enable capabilities intentionally and keep operational automation separate from permission to use a tool.

Distinguish a plausible-sounding answer from one that is actually supported by your business material.

Run the same scenarios against selected models and inspect the evidence behind each answer.

Turn a rough instruction into a clearer draft while retaining control over the final change.

Recognise useful signals such as issue type, sentiment and urgency for support and reporting.

Use supported public-site lookup as a fallback while keeping approved Knowledge current.