Adjusting Tone & Verbosity
Time to complete: 5–10 minutes
Prerequisites: Admin access to set organization-wide instructions
Overview
ConveyorAI always answers by retrieving the relevant sources from your Knowledge Library first. Instructions let you shape how it answers on top of that — the style, the length, and even the content. For example: “Refer to our business as Conveyor, Inc., even if some of our sources refer to us as Conveyor.”
You can set instructions at three levels: for your whole organization, for a single questionnaire, or for a single answer.
🎥 See a demo here! 🎥
How instructions are applied depends on your answer engineAgentic answer engine: ConveyorAI considers your instructions while it answers, not just afterward. They can shape how it approaches a question, including which sources to rely on for specific topics. To see the guidance applied to a specific answer, open the Timeline tab and expand Answer History.
Legacy answer engine: Instructions run after ConveyorAI generates a first draft, as edits to that draft. They work best for wording and style (“Refer to us as Conveyor Inc, not Conveyor”), not for changing how ConveyorAI finds answers (“Look in our SOC 2 first”).
See Agentic Answer Engine to check which engine you’re using.
Organization-wide instructions
Go to Settings → Answering → Pipeline & Instructions. This page also holds your default answering pipeline and whether to import existing answers from uploaded files.
Security questionnaires and RFPs have separate settings, so you can answer each differently. For each one, you can set:
- Verbosity — how long answers should be by default.
- Custom prompt — instructions ConveyorAI follows for every answer. Use them to:
- Influence tone and style (“Refer to our business as Conveyor Inc, and use the first-person plural ‘we’”)
- Influence content (“When the customer asks about our revenue, always reply, ‘We are a private company and do not share this information.’”)
- Apply only when a condition is met (“If the question asks about privacy, add a link to our privacy policy.”)

Set your defaults in Settings → Answering → Pipeline & Instructions.

Set a default verbosity and custom prompt for security questionnaires. RFPs have their own.
Questionnaire-level instructions
When you add a questionnaire, the Style section under AI Settings lets you set a verbosity and custom prompt for that questionnaire only:
- The defaults are your organization-wide settings.
- Your changes apply only to this questionnaire. They won’t affect future questionnaires.
You can also add instructions to an existing questionnaire: bulk select questions and add questionnaire-level instructions, for example “This customer doesn’t use our AI features; answer AI questions as not applicable.”

Set a verbosity and custom prompt for one questionnaire when you add it.
Single-answer adjustments
To polish one answer, select it and click the magic wand icon in the editing toolbar to open the AI Adjust menu:
- Custom Prompt — Tell ConveyorAI exactly how to change the answer.
- Adjust Tone — Change how the answer sounds.
- Adjust Length — Make the answer shorter or longer.
- Regenerate — Run ConveyorAI on the question again.
- Remove Links — Strip links from the answer.

Polish a single answer from the AI Adjust menu.
Common questions
Why didn’t my instruction change the answer?
Open the answer’s Timeline tab and expand Answer History to see whether the guidance was applied. If you’re on the Legacy engine, instructions that change how ConveyorAI finds answers won’t take effect — they only edit the draft.
Can I use different instructions for security questionnaires and RFPs?
Yes. Each has its own verbosity and custom prompt in Pipeline & Instructions.
What’s next?
- Review and Edit Answers — see how instructions shaped an answer
- Agentic Answer Engine — check or change your answering engine
- Troubleshooting inaccurate answers — tell whether a problem came from sources or instructions
Need help? Visit the Troubleshooting guide or contact [email protected].
Updated about 21 hours ago

