### Answer --- The filler words are likely coming from the Natural Filler Words preset in the Agent Handbook. [(1) ]() To disable filler words, open the Agent Handbook in your agent settings (click the Agent Handbook button in the prompt section) and toggle off the Natural Filler Words preset. This preset adds occasional filler words like "um", "uh", and "you know" to make the agent sound more conversational. It's disabled by default for new agents, but if it's been enabled, turning it off will stop the agent from using these fillers. The Natural Filler Words preset is only available for voice agents and adds approximately 100 tokens to each interaction.

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

​ To learn more, see the sources I considered: [[1] Agent Handbook](https://docs.retellai.com/build/agent-handbook) ​

### Answer If the Natural Filler Words preset is already disabled and you're still hearing filler words, the issue may be related to your voice model selection [(1) ]() . Some ElevenLabs voice models are programmed to include filler words regardless of the Natural Filler Words setting . Try switching to a different ElevenLabs model or another TTS provider to see if the behavior changes . Additionally, check your agent's system prompt for any instructions that might be encouraging conversational or natural speech patterns, as these could inadvertently prompt the LLM to generate filler words in its text output before TTS synthesis.

​ To learn more, see the sources I considered: [[1] Agent still saying "Um" despite "Enable Backchanneling" turned off...](https://discord.com/channels/1186424376300679209/1423402635083513967) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer You're correct — if filler words appear in chat testing, it's not the voice provider. [(1) ]() Since the Natural Filler Words preset is disabled and you're seeing fillers in text chat, the issue is coming from your LLM prompt or configuration. Check your agent's general_prompt or state_prompt for any instructions that encourage conversational, natural, or casual language. [(2) ]() [(3) ]() [(4) ]() Even phrases like "speak naturally" or "be conversational" can cause the LLM to generate filler words in its text output. To fix this, review and update your prompt to explicitly instruct the agent to avoid filler words. You can add something like "Do not use filler words such as 'um', 'uh', or 'you know'" directly in your prompt. [(5) ]() If you're using the Retell LLM, you can update the

general_prompt

field via the API or in the dashboard. [(3) ]()

​ To learn more, see the sources I considered: [[1] Agent Handbook](https://docs.retellai.com/build/agent-handbook) [[2] List Retell LLMs](https://docs.retellai.com/api-references/list-retell-llms) [[3] Update Retell LLM](https://docs.retellai.com/api-references/update-retell-llm) [[4] Create Retell LLM](https://docs.retellai.com/api-references/create-retell-llm) [[5] Prompt-Based vs. Conversational Pathway | Retell AI](https://www.retellai.com/blog/prompt-based-vs-conversational-pathways-choosing-the-right-approach) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

Appreciate the feedback! Feel free to ask me again with additional guidance or context.