Processes, automation & AI
I first look at where processes waste unnecessary time, where people are stuck searching, asking and doing manual work – and only then at the tool that will actually help.
This page offers practical orientation, not legal advice.
Where do waiting time, follow-up questions, copy-paste or errors occur? What should become noticeably easier, and for whom?
When rules, inputs and outcomes are clear, classic automation is often cheaper, more stable and easier to trace than AI.
AI becomes useful when language, documents, knowledge or changing cases are involved and fixed rules alone are not enough.
A new chatbot alone does not fix a frustrating workflow. A process automated in the right place and accelerated with AI has a much better chance.
Tool selection is not the starting point. It follows the process, data and risk assessment. The selection below is deliberately not exhaustive, but a practical cross-section of conventional automation, team AI, privacy-conscious usage, local setups and specialist solutions.
We review one concrete workflow and document what should be simplified, automated conventionally or enhanced with AI. You receive a process outline, an initial data and risk assessment, and an actionable next step.
For companies with 10+ employees, loose tool experiments are rarely enough. It takes a small, controlled path into daily work.
Important checkpoint: A pilot starts only after purpose, data classes, risks, responsibilities and the legal framework have been documented. If personal or particularly sensitive data is processed, required contracts and approvals must be in place first.
Which work costs time, energy or quality every day?
Who does what today, with which information, and where does the process get stuck?
Document data types, affected people, protection needs and possible effects. Classify the use under GDPR, the AI Act and any additional industry duties.
Review the legal basis, information duties, deletion periods and responsibilities. Where required, conclude data processing, vendor and subprocessor agreements and adapt privacy notices, terms or customer information with legal support.
Simplify the process, automate with rules or add AI. Define access, logging, data minimization, human approvals and a safe set of test data.
One clear use case, few participants, approved data and fixed review steps. Document benefits, errors and unexpected risks in a traceable way.
Roll out only what works operationally, technically and legally. Enable teams, monitor use and repeat risk and legal reviews when the setup changes.
In the introductory call, we clarify process, data, risks and the smallest sensible starting point – before money goes into a tool or pilot.
Classify the data first
Not every document needs the same protection. What matters is which data is processed and who can access it.
Public information, neutral ideas, empty templates and generally low-risk knowledge work.
Internal material without personal data, heavily abstracted examples and deliberately reduced test data.
Personal data, customer lists, contracts, health-related information and sensitive financial or operational data.
Before use, review the legal basis, contracts or data processing agreement, approvals and any required consent with privacy or legal counsel.
AppSheet and Power Apps can turn spreadsheets and other data sources into useful internal applications. Apps Script and Power Automate automate documents, approvals and recurring workflows – across Google Workspace, Microsoft 365 and connected systems.
These tools are often the better choice when the workflow is clear, rule-based and repeatable. Not every good solution needs a language model.
The right AI tool alone is not enough. Results must be prepared to support decisions and clarify ownership instead of merely producing more output.
Go to reporting & dataBefore the first pilot
AI becomes useful when language, documents, knowledge or changing cases need to be processed. When rules and outcomes are stable, conventional automation is often more reliable and easier to trace.
There is no universal answer. Purpose, legal basis, provider, contracts, access rights and protection requirements must be checked and documented before the pilot. This guidance is not legal advice.
No. Clarify the process, objective and data situation first. Then decide whether existing software, conventional automation, an API or a local system is the right fit.
A focused working session helps: Does the process need simplifying first? Is classic automation enough? Or can AI remove work at one specific point? The goal is a solution that saves time, effort and money in daily operations.
