
What an automation really costs
The quote names one number, for building it. The calculation that decides whether it was worth it has four items, and three only show up after a year.
September 1, 2026 · 10 min readWriting quotes, filing invoices, sorting emails, copying data from A to B: I build automations that take on these and many more tasks. With n8n, Make, Zapier or custom code, connected to the tools you already use.

An enquiry comes in via form or email, AI reads it, checks the data and sorts it, the quote goes out as a PDF, CRM entry included. If you want, you approve the quote first; otherwise that step also runs automatically. All in all it takes about 40 seconds instead of 35 minutes.
Most companies don't lose time to one big thing, but to thirty small ones. Half an hour for a quote, ten minutes for filing, twenty minutes transferring data to accounting. Every week. Per person. Automation targets exactly that: the recurring steps with clear rules.
The task repeats regularly and always runs roughly the same way.
Information travels via copy & paste between tools, emails and spreadsheets.
Mistakes happen because eventually nobody double-checks routine steps.
Something sits untouched because one particular person has no time right now.
An enquiry comes in, the data gets checked, the quote is produced as a PDF and sent out. You just give the final approval.
Form, phone, WhatsApp or a portal: every enquiry lands in the CRM, gets a reply and an owner.
Receipts get read, matched and handed to accounting. Manual filing and follow-up questions disappear.
AI reads along, works out what it's about, suggests a reply and files everything in the right place.
CRM, inventory, shop and spreadsheets talk to each other, so everyone works from the same numbers.
Numbers from every source, every Monday in your inbox. With a short AI-written summary included.
Your case isn't listed? Almost anything that runs by the book today can be automated. Tell me briefly how it works and you'll get an honest read.
Plug in your own numbers: duration × cases × hourly rate. If the result is above €300 a month, it's almost always worth it. On top of the time saved, enquiries stop sitting untouched.
| STEP | BY HAND | AUTOMATED |
|---|---|---|
| Capture the enquiry | Read email, type it in | Automatic, 8 s |
| Check the data | Follow up by email | Follow-up sent automatically |
| Write the quote | Copy template, adjust | PDF from master data |
| File it | Find folder, rename | CRM and Drive at once |
| Follow up | If someone remembers | After 5 days, reliably |
| Total per case | 35 min | 40 s + approval |
A 30-minute call. We go through the process step by step and mark what eats time.
You get it in writing: what gets automated, what it costs and when it goes live. Billed as a fixed price, no timesheets.
First version within 48 hours. We run it alongside your normal workflow until it holds up cleanly.
A short training session and checklist for your team so you can run it yourselves. Or I take on monitoring and further development.
You don't need a new platform that replaces everything. Automations run on n8n, Make or Zapier — self-hosted on your own server for sensitive data. I connect systems that have an interface directly, and build one when there isn't one.
“All of my special requirements were considered and implemented well. The collaboration worked wonderfully.”
One client now receives 25 to 30 enquiries a month that arrive fully captured and routed. Typing things in by hand and hunting through old emails disappears entirely.
Discuss a similar process →The price is set before I start any work.
A first working version you can test yourself.
Workflows, code and docs live in your own accounts.
You talk directly to the developer building the automation.
One clearly-scoped process that runs without you from now on.
Three to five processes that together take a whole area off your plate.
For anyone who'd rather not handle maintenance themselves.
The price range depends on three things: how many steps the process has, how clean the data is, and whether the tools involved have an interface. After the intro call, you get a number that holds.
Not every process should be automated. In these four cases I say so already on the intro call, before any budget flows into a project that won't pay off for you.
If everyone on the team does it differently, you end up automating the chaos too. The process should be clearly defined and followed in practice first, before we automate it. That usually just takes a conversation, not its own project.
Rare tasks don't pay off, and nobody remembers how the automation works. A good checklist is the better solution here.
Where experience and negotiation are involved, I automate at most the preparation. The rest should stay with a human.
Often what's needed first is a clean data foundation or a clear owner on the team. I'll say so openly, so you don't spend the budget for nothing.
By Chris Pump, developer since 2021Last updated: August 2026
Automation is often sold as a huge project: new software, months of rollout, days of consulting. In practice it looks different. The processes that cost small and mid-sized businesses the most time are rarely complicated. There are just too many of them, sitting between systems. That's exactly where automation helps: it connects what's already there and takes over the steps someone used to do by hand.
Both have their place. Where clear rules, fixed formats and the same path always apply, a cleanly built workflow with no AI is entirely enough. But as soon as text needs to be understood, categorized or drafted, language models come into play: summarizing an email, categorizing an enquiry, preparing a quote from a call transcript. In most projects I mix both, because that runs cheaper and more reliably than an AI trying to do everything alone.
You notice the time saved fastest. Reliability matters more in the long run: enquiries don't sit untouched, invoices go out on time, data lives in one place. On top of that comes response speed. Whoever replies to an enquiry within minutes instead of the next day wins business that would otherwise go elsewhere. And because every step is logged, you can see in black and white, for the first time, where the process actually gets stuck.
Before building anything, we clarify which data actually needs to pass through which system at all. Often less is needed than you'd think. When it gets sensitive, we use European providers, self-hosted automation servers, or locally run models. Data processing agreements and deletion periods are settled during the project, not as an afterthought.
At Kasoria you talk to the person who also builds your automation, no account manager and no translation loop between consulting and implementation. I've been developing since 2021 and work with AI every day, so you get an honest read on what actually pays off and what still doesn't run reliably in practice. If an automation doesn't make sense, I'll say that too.

The quote names one number, for building it. The calculation that decides whether it was worth it has four items, and three only show up after a year.
September 1, 2026 · 10 min read
All three connect your tools. The difference doesn't show up in the feature list, it shows up in how they bill, where your data sits, and who still understands the thing two years from now.
August 29, 2026 · 9 min read
Not the most exciting task, but the most boring one. How to work out in an hour which process is actually worth automating, and which one isn't.
August 21, 2026 · 9 min readA single automation is usually live within a few days; you see first results within 48 hours. Bigger bundles take two to four weeks, depending on how quickly I get access and sample data.
No. I build around your existing tools. Only if a system has no interface at all, and none can be added, do we talk about an alternative.
Every process has checkpoints and notifications. Critical steps only run after your approval. If something fails, you see it right away, not only when a customer asks.
Especially then. The smaller the team, the more expensive every hour of manual work is. We start with the process that eats the most time and use it to figure out whether more are worth doing.
Yes. Workflows, code and documentation live in your own accounts. You can keep working on them yourself at any time, or hand them to someone else, without having to ask me for access.
In practice, the tasks that disappear are the ones nobody enjoys: typing things in, filing, chasing up. The work that needs experience and judgment stays with a human. That's exactly what there's more time for then.
Usually yes, but modest: Make and Zapier cost roughly €20 to €80 a month depending on volume, self-hosted n8n around €10 to €20 in server costs. AI steps are billed per request and land in the low single-digit euros for typical volumes. You get the expected running costs in writing before we start.
Either you or I, whichever you prefer. Usually it's a small server with a German provider, running under your own name. I handle updates and monitoring as part of ongoing care, but you always keep access.
Every automation is built to keep running without me, and every process is documented. Critical steps always have a manual fallback. So you're never dependent on me picking up the phone.
Yes, I work remotely only. Projects run over video call and shared access. For the process discovery, a screen share where you show me the workflow once is enough.
Only the systems that appear in the process, ideally with your own user account with limited rights. For sensitive data, we start by working with test data. After the project, access is revoked — that's part of the handover.
Small adjustments are included in ongoing care. Without it, I bill them individually, or you make them yourself: the workflows are deliberately built so texts, recipients and thresholds sit in one place.
30 minutes, free and with no obligation. Afterwards you'll know whether an automation is worth it, what it costs and how fast it runs.
Usually answered the same working day.