"We work with a Personalvermittlung" and "we lease the developer from an agency" sound like they describe the same thing. They do not. Personalvermittlung and Arbeitnehmerüberlassung (employee leasing, sometimes called Zeitarbeit or Arbeitnehmerüberlassung) are two legally distinct models under German labor law, with different contracts, different employer obligations and, for AI and tech hiring specifically, very different fit.
The legal difference, in plain terms
| Personalvermittlung | Arbeitnehmerüberlassung | |
|---|---|---|
| Who employs the person | Your company, directly, from day one | The staffing agency, the whole time |
| Legal basis | Standard employment contract + one-off placement fee | Governed by the AÜG, requires an agency license (Erlaubnis) |
| Duration | Permanent, indefinite by default | Time-bound, tied to the leasing agreement |
| Cost structure | Vermittlungsprovision, typically 20-30% of first-year salary, paid once | Ongoing hourly/monthly markup paid for as long as the person is leased |
| Who manages performance & IP | You, fully, from the first day | Contractually shared, IP and work product terms need explicit handling |
| Best fit | Core, long-term roles you want to own and grow | Short-term capacity, defined projects, seasonal surges |
Why the two get confused constantly
Both models are frequently sold under the umbrella term "Personaldienstleistung," and many agencies in Germany offer both, so the sales conversation can blur the distinction on purpose or by habit. Job boards and LinkedIn posts often use "Personalvermittlung" loosely to describe any third-party staffing arrangement. The confusion is not academic: it changes who your legal employer relationship is with, what rights the worker has, and what happens if the arrangement ends badly.
Which model actually fits AI and tech hiring
AI and senior engineering roles are almost always a better fit for Personalvermittlung, permanent placement, for three reasons. First, IP and code ownership: you want unambiguous ownership of what an ML engineer builds, which is cleanest when they are your direct employee, not a leased worker under a separate employer's contract. Second, team integration: AI work is deeply collaborative and benefits from someone who is fully embedded, not rotating in and out under a leasing term. Third, retention economics: the market for senior AI talent is tight enough that the real risk is losing someone to a competitor, not needing to shed capacity quickly, so the flexibility that leasing buys you is worth less than the stability permanent placement buys you.
When Arbeitnehmerüberlassung actually makes sense
- A clearly time-boxed project (a 3-6 month model migration, a one-off audit) where you genuinely do not want a permanent headcount.
- Bridging capacity while a permanent Personalvermittlung search for the same role runs in parallel.
- Compliance or budget rules in your organization that specifically favor opex over permanent headcount for a defined period.
- Testing fit on a short assignment before converting to a permanent role, some agencies structure this explicitly as "try before you hire."
Questions to ask before signing with either model
- Which contract am I actually signing, and who is the legal employer of record?
- If this is Arbeitnehmerüberlassung, does the agency hold a valid AÜG license, and what happens at the Equal Pay threshold?
- If this is Personalvermittlung, what is the guarantee period if the hire doesn't work out, and is it a refund or a free replacement search?
- Who owns the IP and work product from day one, in writing, not by assumption?