HR
AI for employee onboarding and HR questions
Collect your current HR documents: leave rules, expenses, working hours, equipment and the first-week plan. Build an internal assistant that answers only from these documents and points to the source. Use AI as well to draft onboarding checklists and welcome guides for each role. Keep personal matters such as pay, health or disputes with a person in HR or management.
Effort:
Medium
Time saved:
Roughly 2–4 hours of manager time per new hire; measure it
The problem today
New hires ask the same questions in their first weeks, and managers answer them in between other work. Information is scattered across emails and old files, so every person gets a slightly different onboarding.
How it works, in plain words
Picture a friendly colleague who knows the staff handbook by heart, answers new hires' everyday questions, and sends anything personal straight to HR.
What goes in: your current, approved HR documents: leave rules, expenses, working hours, equipment and the first-week plan for each role.
What the AI does: it searches those documents for the right passage, writes a short answer with the source, and drafts role-based onboarding checklists for managers.
What a person checks: HR or a manager approves the documents and checklists, reviews questions the assistant could not answer, and handles every personal matter.
What comes out: quick, consistent answers for new hires in a tool they already use, and a clear plan for their first day, week and month.
It does not handle pay, health, performance or disputes. It is also only as current as the documents behind it.
How it could work for businesses like yours
Illustrative examples, not client case studies.
A home-goods retailer with 4 stores and 28 staff in Chalkida
Each store manager onboards new sales staff differently, and the same questions about shifts and staff discounts reach the owner. The company would gather its handbook and store procedures, build an internal assistant that cites the document it used, and give managers an AI-drafted first-week checklist to adjust. They would track how many onboarding questions still reach the owner.
A 15-person marketing agency in Rethymno
New hires spend their first week asking where files live, how to log hours and which tools to set up. The agency would turn its scattered notes into one approved wiki and connect an assistant to it. A realistic first target could be a new hire set up on every tool by the end of day two.
The admin team of a 40-person private clinic in Katerini
Reception and admin staff rotate shifts, and newcomers keep asking about rotas, leave requests and equipment rules. The clinic would limit the assistant to approved HR and admin documents only, with no patient data, and send every personal question to the office manager. They would measure how often staff get an answer without interrupting a colleague.
How to do it, step by step
Gather HR policies, the staff handbook and practical information. Check that each document is current and approved, and give it an owner.
Ask AI to draft a role-based onboarding plan for the first day, week and month: what to read, who to meet, which tools to set up. The manager adjusts it.
Build an internal assistant on the approved documents only. Tell it to cite the document and to send people to HR for anything personal or unclear.
Test it with 30–50 real questions from recent hires, in Greek and English if you have international staff. Fix gaps in the documents, not just in the assistant.
Launch it to new hires and their managers with a short note on what it can and cannot answer. Pair it with basic AI literacy training: the AI Act has required businesses using AI to take steps on staff AI literacy since 2 February 2025.
Review the questions every month. The ones it cannot answer show what your documentation is missing.
Tools to consider
Microsoft Copilot Studio: builds an internal assistant on SharePoint HR pages that staff use inside Teams. Caveat: it needs careful permission settings so it cannot surface restricted files.
Custom GPTs in ChatGPT Team or Enterprise: quick to set up on an uploaded staff handbook. Caveat: you must re-upload documents every time a policy changes.
Notion AI: answers questions across your Notion wiki and drafts onboarding pages. Caveat: it is only as good and current as the wiki itself.
Gemini in Google Workspace: drafts onboarding documents and answers questions from Drive files. Caveat: it follows Drive sharing permissions, so loose sharing settings become a risk.
Data protection and GDPR
Limit the assistant's sources to general policies. Do not load personnel files, salaries, medical certificates or performance reviews. Check permissions carefully: an assistant connected to shared drives can surface documents people were never meant to see. Tell staff what is logged when they ask questions, and do not use those logs to monitor individuals. AI used to decide on promotion, termination or task allocation, or to monitor employee performance, is high-risk under the AI Act, so keep this assistant informational. Check with your DPO or legal advisor if you plan to go further.
Mistakes to avoid
Connecting the assistant to all of SharePoint or Drive instead of an approved set of HR documents.
Letting it answer personal questions about pay, leave balances or disputes.
Treating it as a replacement for the manager's conversations in the first week.