Lots of post, little time, high standards of care
Law firms and tax practices work with large volumes of documents: client correspondence, tax assessments, receipts, contracts, pleadings. A considerable part of the work consists of reviewing these documents, assigning them and getting them to the right place. At the same time, the duty of care is high, and mistakes with deadlines can have serious consequences.
AI can ease the load here if it is used correctly: as preparation, not as decision-making. This article shows which processes are suitable, which legal requirements you should be aware of and where the line lies.
Sorting and assigning client post
Whether by email, via a portal or as a scan of paper post, incoming letters have to be assigned to the right client and matter. An AI agent can recognise the sender, subject, reference number and type of document and use them to suggest an assignment.
Typical categories
- Tax assessment, audit notice or enquiry from the tax office
- Letters from the opposing party or a court
- Receipts and documents from clients
- General enquiries that need a response
- Advertising and misdirected post that can be sorted out
If the assignment is uncertain, for example because a reference number is missing, the document goes to a member of staff with a note. The agent does not guess; it asks.
Requesting documents and recognising deadlines
Requesting and assigning documents
In tax practices, chasing missing receipts is part of everyday life. An agent can check which documents are still missing for a month or a year, prepare a reminder to the client and assign incoming receipts to the right period. For further processing, documents can be pre-sorted so that they can be imported cleanly into DATEV, the accounting software used by most German tax advisers.
Recognising deadlines in letters
Tax assessments, court letters and formal requests contain deadlines. The AI can recognise dates and wording that indicate a deadline and suggest an entry for the deadline calendar. Importantly, calculating and monitoring deadlines remains the firm's responsibility. The suggestion is an additional safeguard, not a replacement for your established deadline control.
Drafts and internal research
Standard letters, client newsletters, reminders and answers to frequently asked questions often follow a pattern. An agent can create drafts from your templates, which are then reviewed by a qualified professional. An internal assistant can also answer the team's questions about firm policies or internal templates, citing the source.
A clear line applies to substantive advice. AI does not provide legal or tax advice. It prepares, summarises and suggests. The professional assessment and the responsibility towards the client always lie with the qualified professional.
Professional secrecy and external service providers
In Germany, lawyers and tax advisers are bound by a duty of confidentiality, backed by criminal law in § 203 of the German Criminal Code (StGB). Since a change in the law in 2017, professionals bound by secrecy may involve external service providers where this is necessary for their work and certain conditions are met. For tax advisers this is governed by § 62a of the Tax Advisers Act (StBerG), for lawyers by § 43e of the Federal Lawyers' Act (BRAO).
Put simply, under these provisions the requirements include careful selection of the service provider, a written confidentiality undertaking and limiting access to what is necessary. You should clarify the exact requirements and how they apply to your firm with your professional chamber or a qualified adviser. This article is not legal advice.
- Written confidentiality undertaking from the service provider
- Data processing agreement under Art. 28 GDPR, including technical and organisational measures
- Processing in data centres in Frankfurt am Main
- No training of third-party models on your content
- On request, the model can run in your own infrastructure
- A log of every automated step with timestamp and source
The Scalient Group offers these points as part of the setup. Whether they are sufficient for your particular situation is a decision you make together with your advisers.
A sensible way to start
For law and tax firms, a low-risk, highly repetitive starting point is advisable. Sorting incoming post or requesting missing documents are good candidates: they happen every day, the result is easy to check, and any error will be noticed during further processing anyway.
- Start with one process, not the whole firm
- Initially only suggestions, no automatic actions towards clients or third parties
- Cross-check results for a few weeks, then refine the rules
- Classify and document the use case under the EU AI Act before building
Conclusion
AI can noticeably relieve law firms and tax practices, especially when it comes to sorting, assigning and chasing. The key is a clear division of roles: the AI prepares, the qualified professional decides and takes responsibility. If you also take the requirements for external service providers seriously and make sure processing takes place in Germany, you create the basis for a use of AI that is consistent with professional rules.