AI in everyday work: hand over routines, stay in control
AI in everyday work is already common in Germany, but rarely governed: according to a representative Bitkom survey published on 28 April 2026, 48 percent of employed people use artificial intelligence at work, 8 percent daily and 18 percent at least once a week. 12 percent of users do so without their employer's knowledge, and only 21 percent have attended an AI training course offered by their employer. If you hand routines over to AI, three things keep you in control: a list of approved tasks, fixed review levels per task and a short training session. This guide shows which routines are suitable, what the review levels look like, how much time they free up and which rules apply. As of November 2026.
AI in everyday work: how widespread it is today
The figures show a gap between use and guidance: almost half of employed people use AI, but only 21 percent have used AI training from their employer. The values come from the Bitkom survey of 1,003 people aged 16 and over, interviewed in calendar weeks 8 to 11 of 2026; shares refer to employed respondents.
| Finding | Share |
|---|---|
| Use AI at work | 48 percent |
| Use AI daily | 8 percent |
| Use AI at least once a week, not daily | 18 percent |
| Use AI without the employer's knowledge (among users) | 12 percent |
| Say their job has already changed through AI | 41 percent |
| Have used an AI training course from their employer | 21 percent |
| Employer offers no AI training | 37 percent |
Which tools fit which task is covered in our guide to AI tools for business.
AI in everyday work: which routines you can hand over
Good candidates are tasks that occur often, work with text or tables and whose result a person can check quickly. The AI drafts, the person decides. Unsuitable are tasks where a mistake is costly and hard to spot, such as commitments to customers, figures for the annual accounts or decisions about people.
| Routine | What the AI prepares | What the person checks | Review level |
|---|---|---|---|
| Meeting minutes | Summary, tasks, deadlines | Names, commitments, dates | 2 |
| Customer email draft | Reply based on the request | Tone, prices, commitments | 3 |
| Research on a topic | Overview with sources | Open sources, check figures | 2 |
| Editing a text | Shorten, polish, translate | Meaning and terminology | 1 |
| Analysing a spreadsheet | Totals, anomalies, chart | Calculation on two samples | 2 |
| Job advertisement | Draft from the requirements profile | Non-discrimination, details | 3 |
| Internal knowledge question | Answer from approved documents | Reference in the document | 1 |
| Proposal text | Structure and scope of services | Prices, deadlines, liability | 3 |
AI in everyday work without losing control: three review levels
- Level 1, own review: the result stays internal and is easy to correct. The person using the tool reads it.
- Level 2, review with evidence: the result contains facts, figures or commitments others will reuse. Every figure is checked at its source.
- Level 3, approval by a second person: the result goes outside or has legal or financial consequences. A second person approves it before it leaves the company.
The level is set in the list of approved tasks, not left to individual judgement. The AI does the groundwork, a person always makes the decision.
Worked example: how much time AI in everyday work frees up
An example with assumed values that you replace with your own: an inside sales team of eight spends four hours per person per week on minutes, email drafts and summaries. With AI, pure writing time halves, but reviewing costs a quarter of the previous time.
| Item | Value |
|---|---|
| Writing time so far, eight people at four hours | 32 hours per week |
| With AI: draft (50 percent) plus review (25 percent) | 24 hours per week |
| Saving | 8 hours per week |
| Over 45 working weeks | 360 hours per year |
| At 55 euros internal cost per hour | 19,800 euros per year |
Review time always belongs in the calculation. Licence costs and training stand against the result.
AI in everyday work and the rules: AI Act, works council, data protection
This overview is not legal advice.
- AI literacy under Art. 4 of the AI Act: providers and deployers take measures to support the development of AI literacy among their staff. The Regulation (EU) 2024/1689 prescribes neither a training format nor a certificate. More in our guide to EU AI Act obligations for companies.
- Works council: under Section 90(1) no. 3 of the German Works Constitution Act, the works council must be informed in good time about planned work processes including the use of artificial intelligence. Where tools can evaluate behaviour or performance, co-determination under Section 87(1) no. 6 applies.
- Data protection: personal data belongs only in tools with a data processing agreement. Anyone publishing AI texts should also observe the AI labelling obligation.
Introducing AI in everyday work: a four-week plan
| Week | Step | Result |
|---|---|---|
| 1 | Take stock: who uses which tools for what? | List of tools in use, including private accounts |
| 2 | Set approved tasks and review levels, inform the works council | List with task, tool, data class, review level |
| 3 | 90-minute training per team with real examples | Attendance documented, questions collected |
| 4 | Measure time and refine the list | Saving per task, next tasks |
What such training looks like with your own processes is shown on our page AI training for teams.
Common mistakes with AI in everyday work
- Ban instead of rules: a blanket ban easily pushes use onto private accounts. Already 12 percent of AI users use it without their employer knowing.
- Customer data in the wrong tool: without data classes, contracts end up in services that may use them for training.
- Figures without a cross-check: language models write convincingly even when a number is wrong.
- No measurement: without a baseline, the benefit cannot be shown.
Checklist: AI in everyday work
- List of approved tools with data processing agreements
- List of approved tasks with review level 1, 2 or 3
- Data classes: what may go into which tool
- Works council informed in good time, if there is one
- Training held and attendance recorded
- Contact person for questions named
- Time measured before and after
Questions about AI in everyday work
How many employees use AI in everyday work?
According to Bitkom, 48 percent of employed people in Germany use AI at work, 8 percent daily and 18 percent at least once a week (survey spring 2026).
May I use AI at work without permission?
That depends on your employer's rules. Without approval, data protection and confidentiality breaches are possible, so companies should set clear rules with approved tools.
Must the works council be involved?
Yes, it must be informed in good time about the planned use of AI under Section 90 of the Works Constitution Act. Where tools can evaluate behaviour or performance, it has a right of co-determination.
Is AI training for staff mandatory?
Art. 4 of the AI Act requires measures to support staff AI literacy but prescribes no format and no certificate. A short, documented training session with real examples is the usual route.
Want to introduce AI into your team's everyday work safely and measurably? Go through your task list with us in a short introductory call.
Transparency notice: This article was created with the assistance of artificial intelligence.