Sectors
Four fields of work in which other people’s data fills the working day
What a local AI actually takes on depends on what you work with. For each field this page sets out where the time goes, which three applications pay off first, and how the legal position looks.
Care and nursing homes
Structure care documentation, find documents again, draft letters to relatives and funding bodies — with human oversight.
Where the time goes
Every shift ends in documentation: care reports, handovers, administrative tasks. Findings and letters arrive on paper, relatives and funding bodies are waiting for an answer — and the staff who are supposed to write all of it are missing at the bedside meanwhile.
What helps first
- Turn handover and voice notes taken after a shift into structured drafts for the care records — checked and approved by the nurse
- Match incoming documents to the right resident’s file and make them searchable in the internal archive
- Pre-draft letters to relatives, funding bodies and authorities, and internal administrative tasks, from the records
How we set it up
Local AI can relieve staff particularly in administrative and document-related tasks – structuring notes, preparing documents, searching internal information or producing drafts. Medical or nursing decisions and safety-critical information must not be adopted unchecked from a language model. We design such processes with clear responsibilities and human oversight as a matter of principle. With larger operators, we examine together with IT and data protection staff which requirements for cybersecurity, access, documentation and, where applicable, NIS2 or the respective national transposition need to be taken into account. Because several wards work with it at the same time, the performance class is usually the right choice here.
The legal position
Residents’ files consist almost entirely of health data under Art. 9 GDPR, and nursing staff are bound by a confidentiality duty of their own — in Austria under the Gesundheits- und Krankenpflegegesetz, in Germany, as members of a regulated health profession, likewise through § 203 StGB. On top of that, many residents can no longer give consent themselves; every party processing data outside the building is then one more party to be explained to representatives and supervisory bodies. Local processing reduces additional external data transfers. The processing of personal and in particular health data nevertheless still has to be assessed under data protection law and secured accordingly — best done together with your organisation’s data protection officer.
Law firms
Question the case archive, check pleadings, keep an eye on deadlines.
Where the time goes
The archive grows; the ability to search it does not. Knowledge from closed matters sits in folders nobody opens again, and gets worked out afresh with every similar case.
What helps first
- Check the other side’s pleadings against your own submissions and find contradictions with the page reference
- Assemble earlier matters on the same subject, grouped by outcome
- Go through contracts for deadlines, clauses and departures from the usual wording
How we set it up
As a rule we connect the existing case drive and set up search across the archive as well as drafting of pleadings and correspondence. The matter structure of your filing stays exactly as it is.
The legal position
A lawyer’s duty of confidentiality extends to everything learned in the course of the retainer. In Germany, breaching it is a criminal offence under § 203 StGB; in Austria the professional confidentiality duties of the Rechtsanwaltsordnung apply. If client documents are placed in a cloud outside your own control, at the very least the processing arrangement and the actual access paths have to be examined carefully. If the processing runs on a device in your own rooms, that examination becomes considerably smaller: for the AI itself there is no provider, and anyone with access beyond that is documented in the system planning. How your chamber judges an individual case is a matter to settle with them.
Tax advice and auditing
Pre-sort receipts, bundle queries, find earlier opinions again.
Where the time goes
Receipts arrive in every conceivable format. The same queries to clients cost weeks, and technical points clarified last year can no longer be found.
What helps first
- Read in stacks of receipts, pre-sort them by account, date and amount and report inconsistencies
- Assemble queries to clients automatically from the documents that are missing
- Find earlier opinions on a point across all client files
How we set it up
The usual starting point is pre-entry of receipts together with an automatic list of queries for the client. Search across earlier opinions is added once the archive is connected.
The legal position
Tax advisors are bound by a confidentiality duty of their own — in Germany likewise backed by the criminal sanction in § 203 StGB, in Austria through the Wirtschaftstreuhandberufsgesetz. Client data typically covers the complete financial position of a business: a body of information whose escape causes immediate damage. Local processing keeps it where it belongs. For the professional-law assessment, your chamber is the right address.
Insurance brokers and agencies
Check notifications, compare policy wordings, document advice — health data included.
Where the time goes
Applications and claims are form-heavy and rarely complete. Documenting advice is compulsory but tiresome, and in case of doubt it is kept too short.
What helps first
- Read in claims, check them for completeness and request the missing details
- Compare the policy wordings of different insurers on a single question
- Produce the advice record from the notes of the conversation
How we set it up
Incoming notifications are read in, checked for completeness, and the missing details prepared as a covering letter. The advice record is produced from the notes of the conversation.
The legal position
Applications regularly contain health information and therefore special categories of personal data under Art. 9 GDPR. At the same time the insurance distribution rules require documented advice. Together that makes a body of data to be handled with particular care — and processing it in-house considerably reduces the number of external parties involved.
Your sector is not listed?
The list reflects where we are asked most often so far, not where local AI works. If you work with other people’s data and are not allowed to let it leave the building, the same principle applies — write and tell us what is piling up at your end.
A conversation, not a sales call.
We look at how you work today and work out whether local AI suits your practice. If it does not, we say so — you save weeks of research.
Or simply call: +43 681 10138168