Visitor management: the zone decides, not the job

Starting point

A visit to a meat processing plant touches hygiene, occupational safety, data protection and co-determination all at once. Which briefing is needed, which protective clothing and who has to approve depends on where the visit goes.

The plan is to build it in-house: a digital visitor process for all sites of the group, from the request to the deletion of the data. The project is in the concept phase. This write-up is therefore in the present tense. I describe what exists and claim nothing that does not exist yet.

Decision

First: the zone decides, not the job. A rule set computes from visitor type and zone which briefing, protective clothing, approval and door belong to a visit. The rule set is data, not code.

Second: every record has a legal basis. The modules for briefings and records are versioned. Each names the regulation it rests on. Together they cover accident prevention under DGUV, IFS Food, the QS guideline, EU food hygiene law, the GDPR and co-determination under the German Works Constitution Act.

Third: as little data as possible. The questions on health and animal contact are built so that only “cleared yes or no” is stored. Sign-out and deletion periods run automatically.

Implementation

The process is defined in the concept. The host requests the visit, the responsible manager approves it, the visitor completes the briefings in advance on their phone. At the gate they receive an e-ink badge and an access card.

I verified the connection to the existing access control system instead of assuming it. I read its database and tested its interface live. That turned up a visitor module that had been shipped with the system but never used. It is now an architectural option instead of a parallel world next to it. Sign-out can work without the gate, through the log of door events.

For the badge I evaluated four hardware routes and proposed a pilot with 20 wireless tags at one gate. One hard limit came out of this: no glass in the clean area, because of the IFS glass register.

Result and measurement

That is all there is so far. The process is not built yet. What I can show is the foundation: the rule set stands, every record has its regulation, and the connection to the access control system is verified rather than presumed.

What I would do differently

It is too early to look back. 5 of 8 fundamental decisions are open, and I will not write anything here that has not happened yet. I will fill in this section once the pilot at the gate has run.

Technology

• FastAPI
• React
• PostgreSQL
• E-Ink