AI-Powered Features
Use AI to generate job descriptions, competencies, and interview questions.
Talent Systems uses AI to accelerate your posting setup. These features are optional — you always have full manual control.
AI Job Description
Two options on the JD field:
| Option | What It Does |
|---|---|
| Generate with AI | Creates a full JD from the posting title. Good starting point for common roles. |
| Enhance with AI | Takes your existing JD and improves clarity, structure, and completeness. |
| Regenerate | Once there is a JD, writes a fresh draft. If you've edited the AI's draft (or wrote your own), it asks before replacing your text. |
AI Competency Generation
Click Generate Competencies with AI to get suggestions based on your role. The AI uses:
- Your job description (required)
- Optional recruiter/hiring manager context notes — up to 5,000 characters
- Optional uploaded documents — PDF, DOC, DOCX, or TXT files. Parsed client-side and never stored on our servers.
You review each suggestion and can accept, modify, or reject it.
AI-generated competencies are saved as your organization's custom competencies — not platform competencies — so you can reuse them across future postings.
AI Question Generation
After selecting competencies and depth, click Generate Questions with AI to create a question spine mapped to your selected competencies. Questions are generated from:
- Your competency selections and their behavioral anchors
- The job description
- The selected interview depth
You can edit, reorder (drag-and-drop), or replace any generated question.
AI Practical Briefs
When the AI builds your hiring plan, each practical stage gets a short brief ("Brief for the candidate's task"), a workbench (code editor, SQL, spreadsheet, email, document or diagram) and a time limit. Every candidate in the stage gets a task the AI writes from that brief when they start, fitted to their background (the brief is the same for everyone; the scenario and material can differ).
- Regenerate next to a brief writes a new one for that stage only, in the workbench the stage has. Change the workbench first if you want the task done somewhere else.
- Briefs in the same plan are written to be different tasks, so a candidate never gets the same exercise twice.
- Once candidates have been scored, each stage lists any of its skills that came back Not assessed for some of them ("Terraform: not assessed for 4 of 5 scored candidates"): the stage's work isn't showing that skill. Move the skill to another stage, or change the stage so it can show it.
- Regenerate hiring plan on the "Build the plan" step redraws the skills, where each is measured, the practical briefs and the interview questions. It asks first, because it replaces them.
Every generated practical is checked before the candidate sees it: it must include everything the task refers to (data, logs, emails, files), must not hint at the answer, must be one piece of work that fits the time limit, and starting code must run. A code or SQL practical must also have a checked answer for every test, so every test the candidate runs says passed or failed. A practical that fails is generated again, by a stronger AI model. If it still fails, the candidate gets a library problem of the same workbench instead of an error.
Your Own Practical Tasks (My Library)
When you write a practical task for My Library, you build what the candidate starts with in the same workbench they will use:
| Workbench | What you build |
|---|---|
| Code editor | The starting file and any files beside it (helper code, a data file), as tabs; and whether candidates may add files of their own |
| Spreadsheet | The starting workbook, in a real spreadsheet |
| The inbox: whose mailbox it is, and each email in it (sender, recipients, when it arrived, subject, body) | |
| Diagram | A starting drawing, or nothing to start from a blank canvas |
| Document | Optional starting text |
Draft with AI writes that starting material from the task's title and description (the code and its files, a workbook, an inbox, a drawing), checked the same way as every AI-generated practical. Edit anything it drafts before you save. TSAI library tasks open read only, with a preview of what the candidate starts with.
Hiring Manager Context
Provide additional context that helps the AI interviewer understand the role:
- Context notes — free-text field for recruiter or hiring manager insights (up to 5,000 characters)
- Document upload — role expectations, team wikis, or existing competency frameworks
Documents are parsed entirely in your browser. The extracted text is used to generate competencies or questions but is never stored on our servers.