Forms that fill themselves in
Thousands of previously completed documents serve as examples. Recurring forms are pre-filled automatically, leaving staff to review only what actually needs attention.
AI automation for SMEs · Switzerland (German-speaking) & remote
I do more than advise. I build.
I develop tailored AI modules alongside your day-to-day operations — no forced system change and no disruption to the business. A module only takes over once it has proved itself.
Module board
02 Why things are different now
Custom software used to mean a hard cutover. Now it can prove its value on the side.
Your business does not have to stop in order to improve. New modules can be developed, tested and adopted in a controlled way while your existing operation keeps running.
Then
Major disruption before any value was visible
Custom software meant high costs, a long journey and deep changes to the business. Switching systems required a hard cutover — and put day-to-day operations at risk.
Now
Test first, adopt deliberately
A focused solution can be built cost-effectively alongside your existing system. It is tested in real work and only adopted step by step once it performs. That keeps the risk small without playing down the opportunity.
Modernisation becomes a series of informed decisions, not a bet on one big restart.
03 How I work
One bottleneck. One module. Measurable value.
I do not build a black box designed to replace your entire company at once. I find bottlenecks, build a measurable module for each one, and let it prove that it genuinely helps while your existing operation keeps running.
- 01
Find the bottleneck
We identify where time is being lost, errors occur or work is left undone.
- 02
Define the module
We turn that bottleneck into a clearly bounded task with an understandable goal.
- 03
Test in parallel
The module runs alongside your current operation. Your existing processes stay available.
- 04
Measure value and risk
We make the gains visible, identify the limits and determine whether the module earns its place.
- 05
Adopt it deliberately
The module only becomes part of the core process once it has proved itself.
- 06
Improve without downtime
More modules can be added later, and existing ones can be replaced without bringing work to a halt.
You gain flexibility and planning certainty: controlled modernisation instead of a risky system change.
04 Use cases
Less admin pressure. More time for work that needs people.
This is particularly useful for SMEs under administrative strain — property management companies are a clear example. The best place to start is a recurring process that drains your team every day.
Enquiries routed to the right place
An incoming request, such as a damage report, automatically becomes an email draft for the right service provider at the agreed terms. A person approves it at first. Later, the process can become more autonomous step by step — even when nobody is in the office.
Repetitive admin handled in the background
Recurring tasks are prepared or completed and presented for review. Your team stays in control without having to carry out every step manually.
05 Cost and value
The numbers should still work when the outcome is smaller than hoped.
CHF 80'000–100'000 per year One additional skilled employee
That is the approximate annual cost — if you can find the right person in the first place.
from ~CHF 10'000 One-off investment
Depending on the requirements, AI automation starts around here and can extend into the range of one annual salary.
~1/5 Often a realistic fraction
The investment is often around one fifth of the cost of a new hire. That is an order of magnitude, not a blanket quote.
30 instead of 100 hours/month Calculate conservatively
Even if the module saves thirty hours rather than the hoped-for one hundred, it can still pay for itself — and implementation happened alongside the business, without a hard cutover.
What matters is not an impressive forecast but value you can show: hours saved, a hire avoided or customers you were able to keep serving reliably.
A credible calculation can only come from your actual process. Until then, every figure is an assumption.
06 Services
You do not have to begin with a major project.
The right starting point depends on what you already know and how clearly the bottleneck is defined.
Orientation & training
For curious individuals, self-employed professionals and small teams
An introductory call and a plain-language session give you context, practical guidance and useful resources. You will understand enough to make sound decisions — without being pushed into a large project.
Module consulting + implementation
For SMEs with a specific bottleneck
We analyse the process, define suitable modules and set out the cost, value and risk of each one. You decide what gets built and when. Implementation runs alongside the business.
Continuous optimisation
For companies with roughly 30–50+ employees, or after stage 02
Existing modules are maintained, further opportunities are identified and performance is reviewed regularly. Automation stays useful instead of quietly becoming outdated.
07 A fair split of risk
Until there is proof, I carry part of the risk.
Z4K does not yet have a reference client to point to. That is why suitable early projects follow a straightforward success-based model.
- 01
Costed honestly
The full project is calculated transparently. Scope, assumptions and limits are agreed before work starts.
- 02
Start at ~1/10
Early clients begin with roughly one tenth of the calculated project cost.
- 03
Full amount on delivery
The remaining amount is only due when the contractually defined objectives have been met and can be verified objectively — which forms work, which processes are covered and exactly what counts as complete.
This is not a discount tactic. It is a fair way to share risk while the proof is still being built. I only agree to objectives I am highly confident can be met. Where suitable, existing in-house server capacity can also avoid upfront hardware costs.
08 Principles
Clear limits are part of a good solution.
-
Sometimes good advice is “no”.
I will tell you plainly what AI cannot solve sensibly in your business.
-
No speculative promises.
I do not promise anything I cannot deliver.
-
Modules, not tool sales.
You get tailored, understandable building blocks — not a black box.
-
Value remains visible.
Every solution should show its contribution, whether in time, errors or throughput.
-
Limits are stated.
I am one piece of a larger puzzle. If something sits outside my capabilities, I will say so and point you in the right direction where possible.
-
Strategy, not hype.
No FOMO and no reckless all-in. Do not simply wait, but do not rush in blindly either — decide from sound principles.
09 Jimmy Zarski · Z4K
The change of direction began with an enforced break, not a pitch deck.
Illness forced me to step back. I used that break to choose a new direction before the future chose one for me. I have spent more than 2 years working intensively with AI, paid for plenty of hard lessons and walked away from a startup along the way. What emerged is the system I use today: a factory that builds the next factory.
More than 10 years in film-quality 3D, VFX and content creation shape how I work. Complex ideas are made visual and understandable before you decide. You should be able to see what you are buying, not have to take technical language on faith.
jimmy-portrait Portrait of Jimmy Zarski Placeholder · real visual to follow
Three pillars, one shared foundation
AI automation for SMEs
The work described on this site: understandable modules that solve bottlenecks and prove themselves in live operations.
AI-enhanced video production
The original craft from which the automation work grew. More from ZarskiArt is coming soon.
zarski.art ↗Non-profit work
A cause close to my heart: helping non-profits put AI to genuinely useful work.
lifelion.de ↗ 10 Straight answers
The important objections should be on the table before you get in touch.
01 We already have software. Does it have to go?
No. New modules initially run alongside it and only intervene where doing so is useful and controlled. Once a module has proved itself, you decide whether to adopt it.
02 What if it does not work?
For suitable early projects, the objective is defined contractually and in verifiable terms before work begins. You start at roughly one tenth of the calculated cost; the full amount is only due when the objective is met.
03 Will this replace our staff?
No. AI can prepare or complete 80–90% of a repetitive task. Experienced professionals can immediately sense when an output is wrong. Their knowledge does not become less important — it is applied where it matters most.
04 Are we too early, or already too late?
Early investors spent heavily on immature technology. Later adopters can catch up quickly with the performance available today. What matters is not being first or moving fastest, but starting with a strategy that holds up.
05 Why work with you rather than a large agency?
Because analysis, pattern recognition, implementation and clear visualisation come together in one person. You speak directly with the person building the module. That is also why I deliberately limit the number of clients I work with at once.
06 Do we need to commit to a large project?
No. An orientation or training session may be the right first step. If you already have a clear bottleneck, we start with one well-defined module.
07 How much autonomy does the AI get?
As much as the process can demonstrably handle. At first, it creates drafts for a person to review. Only when quality and limits are understood does the process become more autonomous, step by step.
11 Introductory call
The next step costs nothing but 30 minutes.
Tell me briefly where the work gets stuck. I reply personally, usually within 1–2 business days.