Automation migration with behaviour tests and a rollback
We record how the current automation works, check the same in the new environment and test the rollback. The IT team approves the switch-over.
Automation, paid once
from €3,000 net
Test on your process: €0
You pay 50% after acceptance
How much does migrating an automation to a custom system cost, and when does it pay back?
Short answer
Migrating an automation to a custom system costs from €3,000 net. In our scenario it pays back in 1.5 months. Today, fixing 200 failed runs by hand costs €2,327 a month. With the system, the same work costs €351. The IT team approves every difference and the switch-over. The first stage runs without writing to the production system. The test on your process is free.
Statistics
Teams and AI: statistics 2025
68%
According to Microsoft (2023), 68% of employees say they do not have enough uninterrupted time to focus during the day.
The team gets back about 57 hours of work a month, or 7 working days, and the company €23,710 a year. We count after the time a person spends checking and after the model cost.
200 × 20 min = 66.7 h × €34.90 = €2,327 a month. With the system, a person checks the output in 15% of the old time (€349) and the model costs €2. That leaves €351. The saving is €1,976 a month, so €3,000 pays back in 1.5 months.
Assumptions: 20 minutes to check and fix a failed automation run by hand and checking in 15% of the old time are our assumptions. €34.90 an hour is the EU average hourly labour cost for 2025 according to Eurostat, for companies with 10 or more employees. Model cost: Claude Sonnet 5.5, about 3,000 + 400 tokens per run, at the ECB rate of 1.1205 USD per euro on 22 October 2026.
foap.com customer service: 95% of replies are sent automatically.
Users from several countries write to foap.com, in several languages. The system reads the entire conversation and the terms of that person, and then sends a response in that person's language. Questions about money and exceptions go to the team with a ready draft, so they have to check and send them.
the system sends the answers itself, without human typing
95%
the system sends the answers itself, without human typing
time from the arrival of the e-mail to the ready response (median)
30 s
time from the arrival of the e-mail to the ready response (median)
e-mails that require a response receive a ready-made draft
100%
e-mails that require a response receive a ready-made draft
less work for every 1000 answers (estimate)
47 h
less work for every 1000 answers (estimate)
Data from foap.com production, October 20-21, 2026. Estimate: 3 minutes for handwritten response.
We recorded 336 behaviours of the old environment and checked them in the new one. 7 differences were left for the team to decide, and the rollback to the old version passed its test. Demonstrator on test data.
“We started with my own AI agent. During the training that followed, I learned to build new applications myself. Today I am bringing this way of working with AI into my business. Working with Syntalith gave me exactly the skills and tools I needed.”
You choose the test cases. We anonymise them on request.
They stay in the EU and we delete them after the test.
We sign the NDA and the data processing agreement before we receive anything.
After acceptance you keep the code, the test set and the documentation. You pay for licences and model usage directly to the provider. After acceptance you keep the behaviour tests for later changes.
Questions
How much does an automation migration cost?
From €3,000 net, paid once. After the 5-day test the build takes 2 to 6 weeks. You get a fixed price before the contract: 50% on signing and 50% after acceptance on your own cases.
Will the migration stop the current automation?
No. The first stage runs alongside it, without writing to the production system. The IT team approves the switch-over after the differences are settled.
What if the new version behaves differently?
Every difference goes on a list for the team to decide. We test the rollback to the previous version before the switch-over.
When does it not pay off?
Below 26 failed runs a month the payback takes more than a year. Then it is better to fix the steps that fail most often in your current tool.
What do we need to prepare?
An export of the current automation, run logs from recent months, access to a test environment and the person who knows the flow.
What our OpenAI and Anthropic partnerships mean for you.
OpenAI makes ChatGPT and Anthropic makes Claude. We belong to the partnership programs of both companies, so we work with their tools every day and learn from the creators themselves.
Claude Partner Network
01We will select a tool for the task
We implement both ChatGPT and Claude. We check both on your documents, and you choose the one that makes fewer corrections and saves more time.
02Knowledge straight from the creators
We learn from official OpenAI and Anthropic courses and materials. Your team learns about the features and settings that work today.
03Licenses without our margin
You buy accounts directly from OpenAI or Anthropic and receive an invoice from them. We do not add anything to the license.
04Certified engineers
On the team, we have an AWS certification in machine learning and an Anthropic certification in systems architecture on Claude.
Participation in programs does not mean that OpenAI or Anthropic endorses our services.
customer service
