Skip to main content
Custom report templates replace the default interview report with a holistic evaluation you design yourself. You write the instructions the AI follows, define the exact fields it must return, and build the HTML layout the report renders in. The same templates work for both one-way and conversational interviews.
Custom report templates are included with annual Premium and Elite plans. If you don’t see AI Interview Reports under Templates, check your plan or contact support.

​
Create a Report Template

Templates live in their own library, separate from any single interview, so one template can be reused across many interviews.
1

Open the templates page

Go to Templates, under Account in the sidebar, and open AI Interview Reports. The list shows each template’s Name, Description, Status badge (Active or Archived), and Created date.
2

Name the new template

Enter a Template Name (up to 100 characters, for example “Technical Interview Evaluation”) and a short Description (up to 255 characters) so teammates know when to use it.
3

Fill in the editor tabs

Complete the three required tabs: LLM Prompt, Response Schema, and Jinja Template. Use the fourth tab, Preview, to check the rendered result before saving.
Only Active templates can be attached to an interview. Archive a template to retire it from the dropdown without deleting it. Deleting a template that is currently in use archives it instead, retiring it from the dropdown rather than deleting it outright.

​
Evaluation Prompts

The LLM Prompt tab holds the instructions the AI follows when it evaluates candidate responses. When a template is attached to an interview, this prompt-driven evaluation runs instead of the default evaluator, so the prompt is where you set the bar. Write the prompt the way you would brief a human interviewer. State the competencies to assess, how to weigh them, and what separates a strong answer from a weak one. At the end of your prompt, point to {{response_schema}} so the AI has the schema as context when it writes its evaluation.
Be explicit about every field you expect the AI to fill in. If your schema has a communication_score, tell the prompt what a high or low communication score should mean.

​
Output Schema

The Response Schema tab defines the structure of the AI’s evaluation as a JSON schema. The editor validates that your schema is well-formed JSON, and the AI’s output is constrained to match it. Every field you define here becomes a variable you can place in the report layout. For example:
The schema is what makes reports comparable across candidates. Because every evaluation returns the same fields, you can scan a pipeline of reports and compare using the same criteria.

​
Report Layout and Preview

The Jinja Template tab is where you design the report itself. Write HTML and drop in schema fields as Jinja variables, for example:
The Preview tab renders your layout live. Paste sample data as JSON, such as {"communication_score": 85}, to see exactly how a finished report will look before any candidate has interviewed.
Keep the field names in your Jinja layout identical to the field names in your Response Schema. A variable that does not exist in the schema will never receive a value.

​
Attach a Template to an Interview

Templates are attached per interview. For conversational interviews, pick a template from the Report Template dropdown in the Settings section of the interview’s settings page. For one-way interviews, the template is attached under the template’s own settings instead. Only Active templates appear in the list. Leave the dropdown blank to keep the default report. Once a template is attached, a custom report is generated from it after each interview, and the rendered report appears in place of the default summary on the candidate’s results page.
You can switch back at any time by selecting None in the dropdown, which restores the default report for future interviews.