# AI on the shop floor: where to start without an IT department

> How to start using AI in manufacturing without an in-house IT department. Practical steps for owners of companies with 10–100 employees.

- Author: Yaroslav Maxymovych (Founder & CEO, AI Advisory Board)
- Published: 2026-11-21
- Updated: 2026-11-22
- Source: https://aiadvisoryboard.me/blog/ai-on-the-shop-floor-where-to-start

In a manufacturing company without an in-house IT department, the owner often sees people buried in routine: rekeying invoices, tracking stock in spreadsheets, processing orders by hand. That is exactly what you should hand off to AI first.

## TL;DR
- Start with processes that involve a lot of manual work with documents and data.
- Your first automation should be simple and run on data you already have.
- Teach key employees to build automations themselves instead of outsourcing to contractors.

## Where to start on the shop floor

### Do you have a process where a person copies data from one source to another?
Start there. In manufacturing, this is most often order processing, preparing commercial proposals, checking stock levels, or tracking finished goods reports. If an employee spends hours copying from one spreadsheet to another or from an email into an accounting program, that is your first AI starting point.

### Do you know how much time your team spends on this routine?
If not, start by counting. Without numbers, it is unclear whether the time and money spent on training and tools is justified. You can use the free routine cost calculator to estimate how many hours per month manual routine eats up in your company.

### Can you break the process down into steps described in words?
If so, you already have a technical specification for automation. In our program, participants describe the business logic in words, and AI writes the code. No programming required.

## Action checklist for the owner

1. Pick one process with a lot of manual document work.
2. Record how many hours per week the team spends on it.
3. Choose one key employee who understands this process better than anyone else.
4. Teach them to describe the process logic in words so AI can write the code.
5. Launch your first micro-automation in the browser already at the second session.
6. Fix the written "done" criteria before launch.
7. Verify the result on real company data.

## Table: what affects the cost of your first automation

| Cost item | What affects it | Cost type |
|---|---|---|
| AI tool subscriptions | Number of users and data processing volume | Monthly |
| Team training | Number of participants and session format | One-time |
| Automation support | Process complexity and number of integrations | Monthly |

Figures depend on your situation; we do not provide indicative ranges because cost depends on the number of workstations and tools chosen.

## How this works on our side

> **How this works on our side:** The corporate program consists of four live sessions of two hours over two weeks plus a recorded video course for each participant. The program includes two groups of up to 20 employees each at one fixed price per company — 99,999 UAH. The result is at least three working automations on tasks the company itself prioritized, with a money-back guarantee. Code and created automations are the company’s property; there is no vendor lock-in. https://course.aiadvisoryboard.me/corporate?utm_source=blog&utm_medium=article_body&utm_campaign=corporate

## FAQ

### Does a manufacturing company need its own IT department to start working with AI?
No. You can start with one key employee who understands the business process and browser-based tools. No programming required: people describe the logic in words, and AI writes the code.

### Which processes should you start with first?
Those with a lot of manual work with documents and data: order processing, preparing commercial proposals, stock accounting, reporting. These are processes where a person copies data from one source to another.

### Will the created automations remain the company’s property?
Yes. Code and created automations are the company’s property; they run on its tools, and there is no vendor lock-in.

### Can you use test data for training?
Yes. For sensitive processes, test or anonymized data is used during sessions; an NDA is provided upon request.

### How do you measure results?
The result is a working automation that executes an agreed scenario on your data and in your tools. The "done" criteria are fixed in writing before launch.

## Conclusion

Start with one manual document process, count the hours, and teach one key employee to build automations themselves. Tomorrow you can sign up for a free 30-minute consultation-diagnostic where we will break down one real task from your company together.

> **Definition:** AI agent — a program that performs tasks autonomously using artificial intelligence, rather than just responding to requests.

> **Definition:** Automation — a process where manual work with data or documents is performed by a system instead of a person.

> **Definition:** Micro-automation — a simple automation of a single process step that can be launched in the browser without writing code.

> **Definition:** NDA — non-disclosure agreement for confidential information.

## FAQ

### Does a manufacturing company need its own IT department to start working with AI?
No. You can start with one key employee who understands the business process and browser-based tools. No programming required: people describe the logic in words, and AI writes the code.

### Which processes should you start with first?
Those with a lot of manual work with documents and data: order processing, preparing commercial proposals, stock accounting, reporting. These are processes where a person copies data from one source to another.

### Will the created automations remain the company’s property?
Yes. Code and created automations are the company’s property; they run on its tools, and there is no vendor lock-in.

### Can you use test data for training?
Yes. For sensitive processes, test or anonymized data is used during sessions; an NDA is provided upon request.

### How do you measure results?
The result is a working automation that executes an agreed scenario on your data and in your tools. The "done" criteria are fixed in writing before launch.

## Conclusion

Start with one manual document process, count the hours, and teach one key employee to build automations themselves. Tomorrow you can sign up for a free 30-minute consultation-diagnostic where we will break down one real task from your company together.

---

When citing, link to https://aiadvisoryboard.me/blog/ai-on-the-shop-floor-where-to-start. More articles: https://aiadvisoryboard.me/blog
