Override system prompt on inbound webhook (agent_override)

Currently the inbound webhook agent_override supports overriding voice config, LLM model settings, and various call behavior fields — but not prompt content.

Our use case: we have enterprise customers with customer-specific prompt. Today we have to pre-define these prompt on the agent permanently and are limited to Retell’s basic interpolation.This works but has downsides:

  • Can’t store/manage the prompt in our backend with our custom system
  • Can’t interpolate or preprocess the prompt

Proposed change: Allow agent_override.retell_llm.general_prompt in the inbound webhook response (and create_phone_call API). This would let us dynamically pass an agent prompt. Also for conversational flows ideally.

Thanks!

1 Like

+1 to get this solved. it will help us too.

Definitely something that will be useful.. another perspective as a temp fix if its urgent is passing through prompts as dynamic variables. Then using the dynamic variables to build your prompt.

Whether that is constructing and storing {{system_prompt}} as a DV, passing it in, then in Retell instead of having your prompt there, all it is is {{system_prompt}}.

or If the prompt is generally the same, you just need specific changes, sub those bits out as dynamic variables and feed the changes in.

You could theoritically do this for “node based” prompts too. And “node based transitions”.. though it is quite labourious.

1 Like

That’s exactly what we’re doing right now as a workaround but would be nice to have it officially supported