Set your chatbot's identity, tone, and boundaries

A persona defines how your chatbot introduces itself, writes, and handles its subject area. Your knowledge sources provide the facts; persona and system instructions guide the response style and behavior.

Before you begin

Decide who the assistant represents, what it helps with, and which topics should be redirected to a person. Prepare a short style example and approved wording for any disclaimer. Keep your instructions consistent across Persona, General, and Widget settings.

Example from the Chyt.ai dashboard (September 2026).

Define the identity

1. Open the chatbot's Settings → Persona.
2. In Persona Builder, select Identity.
3. Enter Persona Name, such as “Acme Support Assistant.”
4. Add a Tagline and optional Avatar URL.
5. Write a concise Backstory explaining the assistant's role, audience, and approach.

Describe the assistant accurately. A useful backstory might say it helps customers use your software and explains steps patiently; it does not need an invented biography.

Choose a communication style

1. Select Style.
2. Adjust Vocabulary Level for your audience, from simpler language toward more academic wording.
3. Set Formality.
4. Choose Response Length.
5. Enable Use Emojis only if that matches your brand.
6. Select Traits and choose the qualities you want to emphasize.

These are response preferences. They do not guarantee an exact word count. For a technical upper limit on answer length, use Settings → Model → Advanced response settings.

Set boundaries and save

1. Select Boundaries.
2. Add Allowed Topics, pressing Enter after each topic.
3. Add Prohibited Topics for subjects the assistant should decline or redirect.
4. Enter an optional Custom Disclaimer.
5. Select Save Persona.
6. Open Preview and test an ordinary question, an off-topic request, and a request that needs an escalation.

Persona boundaries are instructions to the model. They are not a substitute for access controls or any separately configured Guardrails.

Align the system prompt

1. Open Settings → General.
2. Review System Prompt.
3. Explain how the assistant should use its knowledge and what to do when information is missing.
4. Remove instructions that conflict with your chosen persona.
5. Select Save Changes and retest.

For example: “Explain answers in plain language. Use our provided knowledge for product and policy details. If a detail is unavailable, say so and offer our published support route.” Avoid embedding changing policy facts in a long prompt when a maintained knowledge source would be clearer.

Reuse a persona

1. Return to Settings → Persona and select Library.
2. Use Save Current Persona to give the current configuration a reusable name.
3. To reuse a saved entry, choose its apply action in Saved Personas.
4. Review the populated configuration.
5. Select Save Persona on the chatbot to persist the applied settings.

Expected result: the chatbot has a saved identity and responds in the intended style during new tests.

Troubleshooting

If the style seems inconsistent, check for conflicting text in the system prompt and Widget → Behavior → Persona & Tone. Use Clear in Preview before retesting so earlier conversation context does not dominate the comparison.

If the tone is right but facts are wrong, update Q&A knowledge or retrieval. If the welcome message or visible widget avatar needs changing, use Widget customization.