Create and save custom prompts for AI Notetaker auto-fill
Ashby now lets teams create and save custom prompts for AI Notetaker feedback form auto-fill, so feedback form drafts automatically follow your team’s preferred style, format, level of detail, and evaluation guidance. Admins can save custom prompts at the org-level, and individuals can save one or more custom prompts that they can reuse.
Set instructions once, use them everywhere
You can now customize AI auto-fill instructions at two levels.
Org-level instructions
Org-level instructions let admins define company-wide formatting and style rules, such as “keep answers under five bullets” or “write in a neutral, professional tone.”
These instructions apply across users and feedback forms by default, giving teams a consistent baseline for how AI-generated feedback drafts should be structured. Admins can manage them in Admin > AI Notetaker.
User-level instructions
User-level instructions let individual interviewers add their own preferences on top of the org-level baseline. Interviewers can save multiple sets of instructions and switch between them depending on the job, interview, or type of feedback they are writing. User-level instructions can be managed from Personal Settings or directly from any feedback form where auto-fill is available.
For example, an interviewer might save one prompt for concise technical feedback and another for more detailed leadership interview notes. They can also set a saved instruction as their default, edit existing instructions, delete instructions they no longer use, or preview saved instructions before selecting them.
In general, user-level instructions are additive to org-level instructions. If the two conflict, such as an org-level instruction asking for bullet points and a user-level instruction asking for paragraphs, the user-level instruction will generally take precedence.
You can also adjust instructions in the moment before filling out a specific feedback form without changing your saved defaults.
Built-in guardrails for org-level instructions
Because org-level instructions apply broadly, Ashby validates them before they are saved.
If an instruction appears to fall outside auto-fill best practices, Ashby will warn admins before saving. This includes instructions that ask AI to make subjective judgments, evaluate or score a candidate, or reference protected or sensitive candidate attributes.
Ashby will also block instructions that appear to manipulate the underlying prompt, such as prompt injection attempts or requests to reveal hidden system instructions.
These guardrails help teams keep AI auto-fill focused on its intended purpose: helping interviewers draft feedback based on their own evaluation, not replacing their judgment.
More relevant answers with more Ashby context
Alongside prompt customization, AI auto-fill now draws on more context already in Ashby, including job descriptions and interview briefing instructions.
That means drafted answers can better reflect what the interviewer was actually asked to evaluate, instead of relying on generic summary text.
Getting Started
Prompt customization for AI Notetaker auto-fill is available for all customers.
Once enabled, admins can manage org-level instructions in Admin > AI Notetaker, and individual users can manage their own instructions from Personal Settings or directly on a feedback form.
Visit our help center for more details.
