What to Automate in Your Company? Which Processes in 2026 (from €3,500 net)
What should you automate first? Start with a process that meets four conditions: it repeats, it has named rules and exceptions, it has volume, and it has a countable cost in hours. Below: the selection method, a map of nine common processes with net ranges, and an honest list of what not to automate. Automating one process starts from €3,500 net.
What should you automate in your company? Start with a process that meets four conditions: it repeats, it has named rules and exceptions, it has volume, and it has a countable cost in hours. Automating one such process starts from €3,500 net. Before you pick a tool, pick the process, and do it on numbers, not on a hunch.
Which processes are even suitable for automation
A good automation candidate meets four conditions at once. Missing even one usually means you should order the process first, not buy a system.
Repeatability. The same work comes back regularly in the same shape: a ticket arrives, you issue an invoice, you update a CRM record. One-off, non-repeating tasks have nothing to automate.
Named rules and exceptions. You can write down what the system should do in the typical case and what to do when something deviates. If the rules live in someone's head and it is always "it depends," get them onto paper first. That is usually 80% of the work before AI even enters the picture.
Volume. The process happens often enough that its cost adds up. Hundreds of cases a month is a different sum from a dozen. Volume decides whether an implementation has anything to pay back.
Countable cost in hours. You know who does it, how often, and at what rate. Without that figure you cannot judge whether automation pays off, and any payback quoted without it would be made up.
If a process meets all four, keep calculating. If it stumbles on rules or volume, fix that first.
Does this process pay off: run it on your own numbers
Payback is not our promise, it is your substitution. Start with what the process costs you today, done manually:
Annual manual process cost =
hours per week on this process
x hourly rate of the people doing it
x 52That one number frames the whole conversation. If the annual cost of manual work is lower than the cost of building and maintaining the automation, we will advise against building it. If it is clearly higher even under cautious assumptions, the process is a candidate. Add the team's quality-control time and a stabilization period after launch, and do not count revenue you cannot measure. The full way to calculate the payback threshold is in the piece on automation cost and ROI.
Process map: what to automate in your company
This is not a ranking, it is a map. Each row is one process that recurs most often at services and trading companies. The key column is "boundary": automation takes the repeatable part, and decisions and exceptions stay with a human. The ranges are for one process taken to production, with integration and a trail.
| Process | What the system takes over | Boundary (what stays with a human) | Typical range (net) | More |
|---|---|---|---|---|
| Shared inbox (office@) | Reads, classifies, drafts replies | Sending sensitive replies and unusual cases | €3,500–8,000 | Inbox triage |
| Invoices and documents (OCR) | Reads, extracts data, enters it into a system | Exception validation and posting to the ledger | €3,500–9,000 | Invoices and OCR |
| Quotes and proposals | Gathers data, assembles a draft quote | Price, terms, and the rep's approval | €4,500–9,000 | Quoting |
| Management report | Pulls data from sources, assembles a recurring report | Interpreting the numbers and decisions | €3,500–8,000 | Management reports |
| Follow-up and CRM | Updates records, watches deadlines, prompts contact | The conversation and the offer itself | €3,500–8,000 | Follow-up and CRM |
| Complaints and returns | Logs the case, classifies it, drafts a decision per policy | Disputed cases and departures from the rules | €6,000–9,000 | Complaints and returns |
| Receivables | Monitors due dates, sends payment reminders | Escalation and hard collection | €3,500–8,000 | Receivables monitoring |
| Meeting notes | Transcribes, extracts decisions and a task list | Priorities and task assignment | €3,500–7,000 | Meeting notes |
| Customer service | Runs repeatable cases within boundaries, escalates the rest | Unusual cases and decisions outside the rules | €6,000–9,000 | Customer service |
The two higher floors are not an accident. Complaints and full customer service rarely fit into a single rule: the process crosses several systems and requires decisions, so the right purchase is an agent that runs the process (from €6,000 net), not simple automation. How to tell one from the other, we explain in the guide on what an AI agent is.
Where to start: one process, not ten
Start with one process, the one with the highest hours-times-rate and the clearest rules. Usually that is the shared inbox, invoice entry, or lead follow-up, because the work is repeatable, the volume is high, and the rules are easy to write down. Do not take the hardest process first, and do not automate several at once.
Run the first launch as a pilot. The system works on real traffic, but a human watches the exceptions and approves whatever goes out. Only once the pilot is stable do you widen the boundaries and take the next process. That order protects you from the most common mistake: deploying everything at once and having no one able to manage it after launch.
How many companies are already automating
For market context, not as a promise: according to Statistics Poland (GUS) data compiled by PIE (December 2025), 8.7% of Polish companies (small, medium, and large combined) used AI in 2025, up from 5.9% a year earlier, with 42% among large firms and 6.1% among small ones. At the same time, an EY Poland report (April 2026, 497 medium and large firms) shows that about half of adopters report disappointment or incomplete ROI, and only 9% have complete data infrastructure. The practical takeaway: the advantage is not "we deployed AI," it is picking the right process and taking it to production.
What NOT to automate
Honestly: there are processes not worth automating, however fashionable it is. Four red flags.
- A process with no rules. If "it depends" means something different every time and no one can write the rules down, there is nothing to code. Order the process first, then come back to AI.
- Low, irregular volume. If a case happens rarely, the cost of building and maintaining it will not pay back even in an optimistic scenario. Manual handling can be cheaper.
- Dirty data at the source. If the input data is incomplete or contradictory, automation only spreads the errors faster. Fix the source first, then automate.
- Decisions about people without oversight. Performance reviews, dismissals, credit or disciplinary calls need a human accountable for the outcome. AI can prepare the material, but it must not make those decisions on its own.
If any of these fits your situation, we will say so plainly before you spend anything.
FAQ
What should a company automate first? One process: the one with the highest hours-times-rate and the cleanest rules. Usually that is a shared inbox, invoice entry, or lead follow-up. Do not automate ten processes at once. Take one, run it through a pilot, then scale. The first step, a free process scan, costs €0.
Which processes can be automated? Processes that repeat, have named rules and exceptions, have volume, and have a countable cost in hours. In practice most often: shared inbox handling, invoices and documents (OCR), quotes and proposals, recurring reports, follow-up and CRM, payment reminders on receivables, and meeting notes turned into tasks.
What is worth automating? Wherever the annual cost of manual work is clearly higher than the cost of building and maintaining the automation. Calculate: hours per week times the hourly rate of the people doing it times 52. If the result tops the build cost, the process is worth it. If not, manual handling can be cheaper, and we will tell you so.
Where do I start with automation? With one process and a pilot, not with a tool. Pick the process with the highest hours-times-rate and the clearest rules, run it on real traffic with a human watching the exceptions, and only after a stable pilot take on more. Start with a free process scan.
How much does automating a process cost? Automating one process starts from €3,500 net, typically €3,500–9,000 depending on integrations and volume. If the process crosses several systems and needs decisions, that is an agent from €6,000 net. The free process scan costs €0.
How to start
Not sure which process? That is what the scan is for. Instead of guessing, calculate a few processes and pick one.
- Book a free process scan and bring two or three processes you suspect.
- For each, prepare: who does it, how many times a month, how long one case takes, which systems are in the path, and where the exceptions appear.
- After the call you get a recommendation: which process to start with, whether it is automation or an agent, and an honest "not worth it yet" if that is what the numbers say.
If you want someone to walk through this with you methodically, that is exactly what the AI process audit is: choosing the right process before you spend money on an implementation.
Book a free process scan | AI process audit | AI automations