FAQ
What matters before we start.
Short answers about AI, data, cost, collaboration, and how we work.
AI, data, and control
Where does my data go when AI is involved?
That is defined before the project starts. Depending on the task, local models, cloud services with a data processing agreement, or a combination are used. You know exactly which services are involved and where data is processed before anything begins.
Can AI make decisions on its own?
Nothing with consequences runs without you. Sending, paying, deleting, pricing, and binding commitments always need your approval. Clearly bounded preparation steps with no real risk, such as sorting or summarizing, can run automatically once they have been checked.
How do you know whether an AI system is dependable enough?
Before daily use we agree how the system should behave on normal cases, missing details, exceptions, and technical failures. We then check exactly that on typical and difficult examples, including whether it stops cleanly and hands the case to the right person. The start is deliberately small. The system takes on more only once the results hold up in daily work.
What does "local" actually mean?
The actual processing runs on your machine, server, or within your network. Which data the system is allowed to send externally is defined deliberately, not assumed by default.
What do you deliberately not automate?
Anything legally, financially, contractually, or personally sensitive is not automated blindly. The system can prepare, flag, or remind. The decision stays with you.
Collaboration and getting started
How does a typical project work?
You describe the process or idea. Together we figure out the smallest useful first step. No months of planning before something usable exists. The first step is scoped small enough to build cleanly and test in practice.
What happens in a first call?
We talk about one concrete workflow: what happens today, where it gets stuck, which tools are involved, and what should run better in the end. After that you get an honest read on whether a small system makes sense, whether the process needs structure first, or whether automation is not worth it right now. Booking through Cal.eu is optional; email is just as fine.
What does a project cost?
Short answer: it depends on how much clarity the process needs. The free first assessment gives you an initial direction. If that's not enough, a workflow and implementation plan usually runs 1,000-2,500 EUR. If you place the follow-up implementation order within 30 days, 50 percent of the plan fee is credited toward the project. A single automation sprint typically costs 2,500-8,000 EUR, a larger operational workflow more like 8,000-25,000 EUR. Optional ongoing support usually starts at 350 EUR per month and can reach about 2,500 EUR per month depending on scope. One thing is fixed: nothing becomes billable until you've seen a concrete proposal.
What do I need to get started?
A specific workflow or idea, two or three real examples, the tools you use today, and a sense of what should work better in the end. No technical knowledge required.
Operations, handover, and support
Do you work with existing tools?
Yes, when it makes sense. Email, spreadsheets, CRM, accounting software, Notion, folder structures, existing data sources. Not everything needs replacing. Often the best solution is one that fits into what is already there.
Does my team need to use AI?
No. In many cases your team just sees a form, a dashboard, an approval step, or an email template. AI works in the background. What is visible is the result, not the technology.
What happens after handover?
You get a documented system you can run yourself. If questions come up or adjustments are needed later, I am reachable. But the goal is always: you are not dependent on me.
Can a first system grow later?
Yes. The first version is deliberately small, but not built as a disposable prototype. If it works in daily use, it can be extended, connected, or operated more cleanly.
Nori Works and how we work
Why one person instead of an agency?
Because you talk directly to the person who builds. Shorter paths, fewer misunderstandings, more honest assessments. For larger projects, I work with a network of experienced freelancers.
What if a project turns out not to be worth it?
Then I say so. If the initial assessment shows the effort does not justify the benefit, or a simpler path exists, that is exactly what you hear. An honest no is better than a project nobody needs.
Do you also build software without AI?
Yes. AI is a tool, not the point. When a problem is better solved with a clean database, a form, or a simple workflow, AI is not needed.
Do you have professional liability insurance?
Yes. Nori Works has professional liability insurance. The insurer is Markel Insurance SE; the policy is managed via exali AG. This is not a quality seal or a guarantee for a specific project, but it is a useful baseline for professional work.
Working together
Got a process that should run better?
Describe it in a few sentences. You get an honest take on whether it can become a working system, and what that would look like.