# InStaff: AI in staffing and workforce services

> How we introduced a GDPR-compliant AI strategy at the staffing platform InStaff and automated its document processing.

Canonical: https://www.waterglass.ai/en/portfolio/instaff/
Language: English (Deutsch: https://www.waterglass.ai/de/portfolio/instaff.md)
Company: Waterglass FlexCo, Börseplatz 1/3/6, 1010 Vienna, Austria · hi@waterglass.ai

InStaff

Kicker: Client

Location: Germany

Sector: HR

Website: https://instaff.jobs

Stat: €25M+

Annual turnover

A German staffing platform for which we introduced and implemented a GDPR-compliant AI strategy.

Max Ferdinand Kunz, CEO & Co-Founder

## Summary

- More than eight endpoints for secure AI document processing, entirely inside the EU
- A custom AI model trained and run on European infrastructure, with a response time under 0.5 seconds
- From almost entirely manual workflows to 55-80% handled by AI per process
- An end to end GDPR-compliant AI strategy introduced and implemented across the operation

InStaff is a German online staffing platform that connects employers with temporary workers for short-term jobs across events, retail, logistics and hospitality – sourcing, contracts, payroll and compliance handled end to end for a pool of more than 100,000 vetted workers. We introduced and implemented a GDPR-compliant AI strategy across the operation: automating document processing to lift the effectiveness and efficiency of the teams. The solution runs on modern AI models, including custom models of our own, so that it holds up to the demands of day-to-day operations. Data protection played a central role from the very start.

## Overview

The first part of the engagement was about automating document processing at InStaff with AI. The second was about building the technical ground for it, one that meets the data-protection requirements of a staffing platform.

The company processes that had been largely manual were handed over to AI step by step, which lifted the effectiveness of the teams. Within a few weeks of going live, depending on the case, up to 55-80% of a process could be handled by AI.

## Context

InStaff places temporary staff on short assignments in events, retail, logistics and hospitality. Every assignment brings paperwork with it: contracts, proofs, timesheets, invoices. With a pool of more than 100,000 vetted workers, that adds up to a volume no team can sensibly keep handling by hand.

Before the engagement, those steps ran almost entirely manually. The model worked, but it showed up in staffing costs. And because personal data is involved, it needed a solution that put data-protection criteria first from the very beginning.

## Secure document processing

We implemented more than eight endpoints covering document processing with AI. Every one of them processes strictly inside the EU and was designed for data-protection compliance from the start. To get there we broke the work structurally into smaller processes, so that individual endpoints could be defined for individual tasks. They fit into existing workflows and could be taken live one after another and improved as they ran.

## A custom model on European infrastructure

For the core of the processing we trained a model of our own. It runs entirely on European infrastructure and answers in under 0.5 seconds. Because it is a small language model (SLM), running it is cost-efficient.

## Outcomes

**From manual to largely automated.** Where the work used to run almost entirely by hand, 55-80% of each process is now handled by AI, supporting the teams so they can concentrate on the complex cases.

**Processing stays inside the EU.** The eight-plus endpoints operate entirely inside the EU, GDPR-compliant and with no detour through a third country. That is how we can be sure data protection is implemented correctly.

**Response times under half a second.** The custom model is fast enough to add no delay to existing workflows. On top of that, running this small language model (SLM) is cost-efficient.

**Still rising.** Because the solution is made of individual endpoints, further tasks keep joining it.

## Engagement model

We developed the AI strategy together with InStaff and then implemented it ourselves: from the individual endpoints through the custom model to the way it sits inside existing workflows. Process automation did not stop at document processing, it was rolled out across large parts of the business.