---
title: "AI onboarding: a 90-day plan for new hires | ekory"
description: "How to onboard a new hire in 90 days: what to prepare before day one, the first week, the first month. Where an AI agent helps, and where it does not."
url: https://ekory.ai/blog/onboarding-with-ai-first-90-days
language: en
italian: https://ekory.ai/it/blog/onboarding-with-ai-first-90-days.md
updated: 2026-10-07
---

Guide 7 October 2026 10 min read

# AI onboarding: the first 90 days, week by week

A new hire takes about four months on average to become fully productive and absorbs around a hundred hours of colleagues’ time. Here is how to organize the first ninety days, what to prepare before day one, and where an AI agent working on the company’s documents lightens everyone’s load.

- **4 months** average time for a skilled new hire to become fully productive
- **100 hours** of colleagues’ time absorbed by each new hire
- **12%** of employees strongly agree their organization onboards new people well (Gallup)

## Key takeaways

- **A 90-day onboarding plan has four phases**: preparation before day one, a first week to find one’s bearings, a first month to learn the job, and days 30 to 90 to work independently.
- **Before day one** you need working access, the documents of the role gathered in one place, a go-to colleague with time set aside, and the mandatory health and safety training in the calendar, which Italian law requires at the start of employment.
- **An AI agent working on company documents** answers the new hire’s questions with the source, turns existing materials into a course with lessons, podcasts and quizzes, and shows the manager who has completed it.
- **People stay irreplaceable** for relationships, judgment on difficult cases, feedback and evaluation: AI removes repeated questions, not the bond with colleagues.
- **Measure it this way**: days to the first task done alone, repeated questions to the most experienced colleague, courses completed, and a review at 30, 60 and 90 days.

## Why the first 90 days decide onboarding

A new hire does not perform like their colleagues for months. Employer surveys in Germany and Switzerland put it at three to four months for a skilled worker, and each new hire absorbs about a hundred hours of colleagues’ time on average. In our [study on the cost of onboarding](https://ekory.ai/blog/the-cost-of-starting-from-scratch) we estimated that, for an Italian company, all this is worth between €5,300 and €9,300 per person.

The first ninety days cover most of that period, and they often coincide with the probation period. In Italy it can last up to six months for open-ended contracts, unless collective agreements set a shorter term. For fixed-term contracts it is one day of actual work for every fifteen calendar days, with a minimum of two days and a maximum of fifteen for contracts up to six months, or thirty for contracts between six and twelve months, unless collective agreements are more favorable. It is the time in which the company decides whether the person is the right one, and the person decides whether the company is.

A good onboarding plan is not a document: it is a sequence of things done at the right moment. Research sums up the goal as three states to reach as soon as possible: understanding one’s role, feeling able to do it, and feeling accepted by colleagues. Everything that follows serves one of the three.

## What to prepare before day one

Every day a new hire spends waiting for access or asking where things are is paid with no output. Preparation happens the week before.

- **Access and tools.** Accounts, email, folders, business software: only the ones the role needs, ready on the morning of day one.
- **The documents of the role, in one place.** Procedures, price lists, technical sheets, templates, the answers to the most frequent questions. You do not need to write new ones: you need to gather the ones that exist and remove outdated versions.
- **A go-to colleague, with real hours.** A Swiss employer survey puts onboarding at about a hundred hours of colleagues’ time, so plan them: an hour a day blocked in the calendar in the first week, a few hours a week after that. Without dedicated time, shadowing turns into a constant interruption.
- **Mandatory training in the calendar.** Health and safety training must take place at the start of employment (article 37 of Legislative Decree 81/2008): put it in the first days, not whenever it happens.
- **A written plan for the first two weeks**: who they meet, what they read, which first task they handle.
- **A small, real first task**, to be completed within the first week.

Where AI helps: an AI agent connected to the documents of the role can be ready on day one. The new hire sees only what they are allowed to see, and finds an answer with its source instead of a folder to explore.

## The first week: finding one’s bearings

The goal of the first week is not to produce: it is to know whom to ask, where things are and how work gets done.

1. **Day 1.** A welcome, a round of the people they will actually work with, tools that work, lunch with the go-to colleague. By the evening the new hire should know what they will do tomorrow.
2. **Days 2 and 3.** Shadowing on real tasks, with permission to ask every question. Health and safety training, if it has not happened yet.
3. **Days 4 and 5.** The small first task done independently, then reviewed together. A short check-in with the manager at the end of the week: what is clear, what is not.

Where AI helps: first-week questions are almost always the same. How to file an expense claim, which price list applies, who approves a discount. An AI agent answers them citing the document and page, so the experienced colleague only answers the ones that need their judgment. Questions with no answer in the documents make a useful list: they show what is missing.

## The first month: learning the job

In the first month the new hire moves from watching to doing. Three things are needed: content to study, real work to do, and regular feedback.

- **A training path for the role**, built on the company’s own materials: products, customers, procedures, typical mistakes.
- **Growing responsibility**: from tasks reviewed one by one to tasks checked by sampling.
- **A weekly meeting** with the manager and frequent meetings with the go-to colleague. In Microsoft’s program, the share of new hires who say their onboarding buddy helped them become productive quickly rises from 56% among those who met once to 97% among those who met more than eight times in the first 90 days.

Where AI helps: preparing a course is the job that usually does not get done, because it takes weeks from people who already have a full-time job. With an AI agent, the documents the company already has become a course with lessons, two-voice podcasts to listen to in the car, and quizzes to check what is clear. The manager assigns the course and sees who has completed it; the expert reviews and corrects instead of writing from scratch.

## Days 30 to 90: working independently

In the second and third months the new hire works more and more on their own. The risk changes: it is no longer the questions, it is the exceptions. The customer with special terms, the case the procedure does not foresee.

1. **At 30 days**: a formal check-in. What they can do alone, what not yet, what was missing in their onboarding.
2. **Between 30 and 60 days**: measurable goals for the role, with the go-to colleague available for difficult cases.
3. **At 60 days**: a review of goals and relationships. Does the new hire know the people they work with?
4. **At 90 days**: an assessment of onboarding from both sides, and a plan for the following months.

Where AI helps: exceptions are where a company’s knowledge is least written down. Every time a colleague answers a new question, that answer can become a document the AI agent will use next time. In ninety days a new hire asks dozens of questions: if every answer stays, the next hire starts further ahead.

## Where an AI agent helps, and where people are needed

Who does what in onboarding

| Activity | AI agent | People |
| --- | --- | --- |
| Answering questions on procedures, products, price lists | Answers straight away, with the source | Check the answers and fix the documents |
| Preparing training for the role | Turns documents into lessons, podcasts and quizzes | Choose the right materials and review |
| Following progress | Shows who has completed the course | Give feedback and decide the next steps |
| Difficult cases and exceptions | Finds the written precedents | Decide |
| Relationships and culture | No role | Welcome, introduce, include |
| Evaluation and probation | No role | Evaluate and decide |

Three caveats. First: an AI agent is as good as the documents it reads; if procedures are outdated it will answer with outdated procedures, which is why the cited source matters. Second: it must see only what the new hire is allowed to see, not the whole company. Third: if you put an artificial intelligence tool in employees’ hands, the AI Act requires measures to ensure a sufficient level of AI literacy among the people using it; we wrote about it in [You are already an AI deployer](https://ekory.ai/blog/you-are-already-an-ai-deployer).

What AI does not replace is the point research insists on most: acceptance by colleagues. A new hire who no longer has to ask the trivial things has more time for the conversations that matter, not less.

This is how ekory works: Ekos, the AI agents, read the documents you choose, answer citing document and page, and from the same materials prepare the course for the role, with managers seeing who has completed it. Each person sees only what they are allowed to see. See how it works for [fast-growing companies](https://ekory.ai/scaleups) and on the [product page](https://ekory.ai/product).

## How to tell whether onboarding works

- **Time to the first independent task**: how many days pass before the new hire completes a typical task on their own.
- **Repeated questions to the most experienced colleague**: if they drop week after week, the documents are doing their job.
- **Courses completed and quizzes passed**, to spot what is unclear before it turns into a mistake.
- **The 90-day review**, from both sides, and how many people are still there after the first year.

The benchmark to beat is low: according to Gallup, only 12% of employees strongly agree that their company is excellent at onboarding new people.

## The 90-day plan in one table

Onboarding plan for the first 90 days

| When | Goal | What the company prepares | Where the AI agent helps |
| --- | --- | --- | --- |
| The week before | Everything ready on arrival | Access, documents of the role, go-to colleague, safety training scheduled | Connected to the documents of the role, with the right permissions |
| Week 1 | Find one’s bearings | Welcome, shadowing, a small first task | Answers practical questions with the source |
| Weeks 2 to 4 | Learn the job | Training path, weekly meetings | Prepares the course with lessons, podcasts and quizzes |
| Days 30 to 60 | Work with light supervision | Measurable goals, difficult cases with the colleague | Finds precedents; new answers become documents |
| Days 60 to 90 | Work independently | Reviews at 60 and 90 days | Shows what has been completed and what has not |

## Frequently asked questions

### What is a 90-day onboarding plan?

It is a program that organizes a new hire’s first three months, with goals for the first week, the first month and days 30 to 90. It helps the person work independently sooner and lets the company see early on whether onboarding is working.

### How long does onboarding a new employee take?

For a skilled worker it takes three to four months on average to perform like colleagues, according to German and Swiss employer surveys. People who change sector or start their first job can take two or three times as long.

### How do you train a new employee with AI?

By connecting an AI agent to the company’s documents: it answers questions citing the source and turns the same materials into a course with lessons, podcasts and quizzes. The manager assigns the course and sees who has completed it, while the expert reviews instead of writing from scratch.

### Can AI replace shadowing a colleague?

No. It can take repeated questions and the preparation of materials off the colleague’s plate, but relationships, judgment on difficult cases, feedback and evaluation stay with people. Research identifies acceptance by colleagues as one of the factors that decide whether onboarding works.

### What does Italian law require for training new hires?

Health and safety training must take place at the start of employment, as set by article 37 of Legislative Decree 81/2008. If the new hire uses artificial intelligence tools, the AI Act also requires measures to ensure a sufficient level of AI literacy.

### Which documents do you need to onboard a new hire?

The ones that answer the questions of the role: procedures, price lists, technical sheets, templates, contacts and the answers given in the past. You do not need new documents to start: you need the existing ones, gathered in one place, up to date and easy to ask.

## Sources

1. ekory, [_The cost of starting from scratch_](https://ekory.ai/blog/the-cost-of-starting-from-scratch), October 2026.
2. T. N. Bauer et al., [“Newcomer adjustment during organizational socialization: A meta-analytic review of antecedents, outcomes, and methods”](https://doi.org/10.1037/0021-9010.92.3.707), _Journal of Applied Psychology_, 92(3), 2007.
3. T. N. Bauer, [_Onboarding New Employees: Maximizing Success_](https://studentaffairs.uga.edu/wp-content/uploads/2021/08/Onboarding-New-Employees.pdf), SHRM Foundation, 2010 (copy hosted by the University of Georgia).
4. Microsoft WorkLab, [“Strategies for Onboarding in a Hybrid World”](https://www.microsoft.com/en-us/worklab/strategies-for-onboarding-in-a-hybrid-world), 29 June 2021.
5. Gallup, [“Why the Onboarding Experience Is Key for Retention”](https://www.gallup.com/workplace/235121/why-onboarding-experience-key-retention.aspx), 16 May 2018.
6. Legislative Decree 81/2008, [article 37](https://www.normattiva.it/uri-res/N2Ls?urn:nir:stato:decreto.legislativo:2008-04-09;81~art37), Normattiva.
7. Legislative Decree 104/2022, [article 7](https://www.normattiva.it/uri-res/N2Ls?urn:nir:stato:decreto.legislativo:2022-06-27;104~art7), as amended by Law 203/2024, [article 13](https://www.normattiva.it/uri-res/N2Ls?urn:nir:stato:legge:2024-12-13;203~art13), Normattiva.
8. [Regulation (EU) 2024/1689](https://eur-lex.europa.eu/eli/reg/2024/1689/oj) (AI Act), article 4, EUR-Lex.

## More research and articles.

- [AI for publishers and literary agencies: rights, catalogs, royalty statements and manuscripts](https://ekory.ai/blog/ai-for-publishers-and-literary-agencies): Industries 7 October 2026 8,645 translation rights sold abroad in 2025, three publishers in four already using AI: where it helps with contracts and royalties, and where the copyright line runs.

- [AI for accountants and labor consultants: practical uses, limits and rules (2026)](https://ekory.ai/blog/ai-for-accountants-and-payroll-consultants): Industries 29 September 2026 Since 10 October 2025, clients must be told which AI you use. What it can do in a firm, what stays with the professional, which rules apply.

- [All research and articles.](https://ekory.ai/blog): Resources Data, regulation and cases, in English and Italian.

## Tell us about a process you still do by hand. We'll show it to you on your own materials.

For the first call, your website or a catalog is all we need.