Five levels of
AI adoption
AI at work, from kindergarten to professor: two company stories, one study, and examples for every role.
Hands up
- Who used AI this week?
- Who used it for work?
- Who used it for a task that took more than 30 minutes before?
AI takes the repetitive half of your week first
Developers
Reading old code, writing tests, documentation.Support and helpdesk
First replies, ticket summaries, the same question for the tenth time.Admins
Scripts, reading logs, runbooks, incident write-ups.The time you get back goes to the part of the job people are hired for: judgment.
One European tech company
~80%
faster delivery, in Java and in SAP
AI across software development, the SAP team included. If it works on their ABAP, it can work on yours.
The company’s engineering leadership, as told to Marian Kamenistak. Name withheld.
One global tech company
Everyone works with Claude Desktop
It is expected from every employee, in every department. People stop asking whether to use AI and start comparing how they use it.
A leader at the company, as told to Marian Kamenistak. Name withheld.
5,179 support agents got an AI assistant
tickets solved per hour, all agents
tickets solved per hour, the newest agents
With AI, agents two months in worked like agents six months in without it.
Brynjolfsson, Li, Raymond: Generative AI at Work. NBER working paper 31161, 2023. Published in the Quarterly Journal of Economics, 2025
Five levels of AI at work, from kindergarten to professor
At every level: ask for a plan or questions first, then let it work.
Level 1: Kindergarten
Ask
You ask, it answers, you copy. A smarter search.
Developers
“Explain this stack trace.” “What does this SELECT do?”Support and helpdesk
“Rewrite this answer so a warehouse colleague understands it.”Admins
“What does this PowerShell command do before I run it?”Everyone in the company
Ask before you open a ticket: “Excel shows #REF!, why?”Level 2: Elementary school
Brief
You give it your context: the ticket, the log, the spec, your rules.
Developers
Paste the spec and the class. Ask for a plan and open questions before any code.Support and helpdesk
Paste the ticket and the knowledge-base article. Get a reply draft in Czech, English or German.Admins
Paste the log and the config. Ask for three possible causes and how to check each.Everyone in the company
Paste the error and what you did. The helpdesk can solve your ticket on first read.Level 3: College
Repeat
The prompt you write every week becomes a saved template or skill.
Developers
A team CLAUDE.md with your naming and rules. Saved skills: code review, unit tests, documentation.Support and helpdesk
Saved prompts: ticket summary, standard replies, handover note for the next shift.Admins
Runbooks as saved prompts: new laptop setup, patch night checklist, incident write-up.Everyone in the company
Your own saved helpers: an email in German, meeting notes into tasks, the weekly report.Level 4: University
Connect
AI reads and acts in your tools: tickets, monitoring, SAP, Jira.
Developers
AI reads the repository, the test results and the tickets. In SAP: the DEV system, ATC findings, ABAP Unit runs.Support and helpdesk
AI reads the queue and the knowledge base. It suggests category, priority and duplicate tickets.Admins
AI reads monitoring and logs, builds the incident timeline, prepares the script for your review.Everyone in the company
A company assistant that knows the IT guides and the status of your request.Level 5: Professor
Supervise
AI runs the process end to end. You set the rules and approve.
Developers
AI drafts the change, runs the tests and checks, hands it over for review. A person approves the release.Support and helpdesk
AI solves how-to tickets end to end. Anything with access rights or money goes to a person.Admins
AI runs routine checks every night, reports drift, fixes from an approved list. You approve the rest.Everyone in the company
“New colleague starts Monday.” Access, laptop and accounts get requested. You confirm.Where are you today?
- Pick your level, kindergarten to professor.
- Tell the person next to you, with one example.
- Goal: one level up in 30 days.
Run experiments.
Play with Claude.
Otherwise you will not be needed in the future.
It is your own choice.
One task,
one level up
Pick one task you did three times last week. On Monday, do it with AI one level higher than usual. Write down the minutes it took.
Where to go from here
Your company, not only you
The same five levels for a whole engineering organisation, with the signs of each level.AI in an SAP team
The slides of a full workshop day: install guide, ten AI skills, the exact prompts.Learn it with a lecturer
Workshops and the online AI Program of the Engineering Leaders Academy.The article behind these slides: How to use AI at work: 5 levels, kindergarten to professor
Slides by Marian Kamenistak, Engineering Leaders Community. Free to reuse with a link to this page.
Apply the five levels to your role
- Scan the code to open these slides.
- Find your role on each level, 1 to 5.
- Improve: one level up, then the next.