All articles
Automation
Support, HR and back office9 min

Employee onboarding: AI handles first-week questions so the mentor can focus on what matters

A new hire's first week tends to follow the same script. They hesitate to interrupt their mentor again, dig through old messages for the CRM password, have no idea who signs off their building pass, and by Friday they have formed a view of the company based on how hard it was to get a straight answer. The mentor, meanwhile, is trying to do their own job between dozens of small "where do I find…" questions. This article covers how to put an AI assistant next to a new employee specifically for week one: which questions it handles on which day, when it should message first, where it must hand over to a person, and how to prepare its material in a few working days rather than a quarter.

What new hires ask in their first five days

First-week questions are predictable, and that is exactly what makes them a good fit for AI. They also shift day by day: first logistics and access, then tools, then processes and people. Ask your last two or three hires what they asked, and you will get something close to this:

DayTypical questionsWhere the answer lives
1 — logistics and accessWhat time to arrive, where people eat, how to join the Wi-Fi, who takes my documents for HR, when the laptop arrivesOffice guide, contact list, IT checklist
2 — toolsHow to log in to the CRM, where the email templates are, which chat is for whatSoftware guides, map of work chats
3 — processesHow a deal moves from enquiry to payment, who approves a discount, how refunds workTeam procedures, process diagram
4 — people and rolesWho owns purchasing, who to ask about deliveries, who the person everyone keeps mentioning actually isOrg chart with areas of responsibility
5 — wrap-upWhat is expected of me by month end, how probation is assessed, where to see my tasksOnboarding plan, agreements with the manager

The first four rows are reference questions: the answer is the same for everyone and is written down somewhere, or should be. The fifth is a conversation with the manager, and the AI's only job there is to remind both sides that the meeting is in the plan.

How an onboarding assistant differs from a help desk bot

An internal help desk bot waits to be asked; we covered that model in our piece on an internal help desk built on a knowledge base. For a new hire that is not enough. They often do not know what to ask and feel awkward asking anything that sounds obvious. So an onboarding assistant runs on a schedule and speaks first:

  • Day one, morning: a short welcome, the plan for the day, the mentor's name and three things to get done before lunch.
  • Day one, end of day: "Did all your access come through?" with buttons for email, CRM, work chats and building pass.
  • Days two and three: one short tip a day about the process the new hire is starting on, with a link to the guide.
  • Day four: who owns what in the team, and which questions go to whom.
  • Day five: three feedback questions: what was unclear, what you could not find, what you wish you had known earlier.

If the new hire reports that CRM access still has not arrived, the assistant does not offer sympathy. It opens a task for whoever handles access and lets the mentor know. That is a case where automation saves not minutes but an entire lost day.

Example: a sales hire on day two

An illustrative conversation in a messaging app. The mentor has gone home; the new hire messages the assistant in the evening:

New hire: I'm handling inbound enquiries tomorrow. A customer is asking about paying in instalments. Do we offer that?

Assistant: Yes. Under the sales procedure, instalments go through a partner bank; you only take the request and pass it to the processing team (section 4, "Payment methods"). Don't quote terms yourself, the bank confirms those. There is a reply template under "Templates → Payment".

New hire: And if they ask for a discount for paying cash?

Assistant: Discounts beyond the price list are approved by the head of sales, see section 5.1. I've added this to the list for tomorrow's session with your mentor, since it's worth going through real examples of how the team usually responds.

The assistant did not invent a discount policy, and it did not leave the new hire stuck either. It gave the rule and moved the judgement call to a person. Section and folder names here are placeholders; yours will differ.

Who owns what during week one

TaskWhat the AI doesWhat stays with people
Logistics, access, contactsAnswers from the office guide, checks everything was issued, raises tasks for anything missingIT grants the access
Using the softwareWalks through the guide step by step, in whatever language the hire asks inMentor demonstrates on a live task
Processes and rulesQuotes the procedure and says who approves exceptionsManager explains why the rule exists
Employment paperworkSays which documents to bring and to whomHR processes them and handles anything personal
Expectations and probationReminds about the meeting and collects questions for itManager sets goals and gives feedback
Culture and unwritten rulesNothing, because none of it is written downThe team

Building the first-week pack

A common mistake is to start by documenting every process in the company. Onboarding does not need that. A new hire uses a small slice of company knowledge in week one, and that slice can be assembled quickly.

  1. Interview your most recent hires. Twenty minutes each with two or three people who joined in the last six months: what they asked, what took longest to find, what they learned too late.
  2. List the questions by day. Use the table above as a skeleton. You will usually end up with a few dozen questions, many of them shared across teams.
  3. Write short answers. A paragraph each, with a file link where one exists and the name of the topic owner. If the answer is nowhere, that is a finding in itself: the rule lives in someone's head. We hit exactly this when moving our own processes into one system (case study).
  4. Plan the check-ins. Five to seven proactive messages over the week is plenty; more starts to feel like nagging.
  5. Split by role. Sales, warehouse and finance need different guides. The shared layer is the same for everyone; the assistant only answers from what the new hire's role is allowed to see.
  6. Test it on your next hire. On Friday, sit down with the mentor, review what the assistant could not answer, and fill the gaps for the next person.

If your onboarding really starts before day one, with screening and interviews, see how AI helps with first-pass screening of applications.

What the manager sees on Friday, and the limits

By the end of the week you have a list of everything the new hire asked. It shows which topics caused friction, which guides are out of date, and which questions had no answer at all. Use it to improve onboarding and to brief the mentor, not to assess the employee, and say so on day one. Otherwise people stop asking.

  • The assistant is only as good as the pack. Old screenshots in a guide mean it will confidently describe a button that no longer exists. Every answer needs an owner and a last-updated date.
  • It does not transmit culture. How the team disagrees, who you can message at the weekend, what "urgent" means to this manager: people learn that from people.
  • It is not worth it if you hire once a year. Update the guide and give the mentor time instead. It pays off when people join regularly or across several locations.
  • Personal matters go to HR only. Pay, sick leave and family circumstances should never pass through the assistant's shared log.

More on internal processes is in our Operations section.

FAQ

How is this different from a PDF welcome guide?

A guide gets read once on day one and forgotten. The assistant answers when the question actually comes up, searches all documents at once, checks in on schedule and notices when something has not been issued. The questions it cannot answer tell you what the guide is missing.

Which languages can it handle?

It replies in the language it is asked in, even when the source guide exists in only one language, and links back to the original document. That matters for mixed teams, whether in Dubai, London or Tashkent.

Can the same assistant stay on after week one?

Yes. It usually turns into an internal help desk for all staff: the check-in schedule switches off and the question-and-answer base keeps growing.

How much does it cost?

It depends on the number of roles, the volume of documents and whether it needs to connect to your CRM or access management. The first conversation is to scope it for your company: get in touch, and read how we approach these projects on our AI agents page.

Subscribe to AI Insights

Case studies, tips and AI automation trends — weekly

Want similar results?

Get a free consultation and ROI calculation for your company

Free Consultation

Related Articles

Automation

An internal help desk on your knowledge base: moving staff questions out of the group chat without losing any

Which employee questions AI can close from your knowledge base, where the rest go, and how every unanswered question becomes a new policy article.

Automation

Supplier invoices without manual keying: how AI reads bills and delivery notes while your accountant only reviews

Which documents AI reads reliably, which checks must run after extraction, and why posting and payment still stay with your finance team.

HR

Job applications from job boards, messengers and your website: how to merge them into one queue and let AI do the first screen

A practical setup for high-volume hiring: one candidate record, knockout questions in a messenger bot, replies in the candidate's language and clear limits where HR decides.

Guide

How an AI Chatbot Replaces First-Line Support

The architecture explained: what a support chatbot is made of, which enquiries it genuinely closes and which it does not, how to calculate payback on your own numbers, and the snags that surface in week two.

HR

AI Agents for HR: From CV Screening to Onboarding

Where HR actually loses time and which four tasks get automated first: screening applications, coordinating interviews, onboarding, and answering routine employee questions. Including what should not be automated at all.

Automation

AI Agents for Private Schools and Learning Centres: Admissions Desk, Parent Enquiries and Timetable Changes

Three streams that eat up a school administrator's day — parent questions, trial lesson bookings and timetable changes. What you can hand to AI and what you shouldn't.

Employee onboarding with AI: the first week | UNIKA