Process first, technology second
Automation is not a product you buy; it is a way of looking at work. If a process is a mess, automating it just makes the mess faster. That is why we start by mapping the procedure with the people who run it every day.
I look for three signals: repetitive rule-based steps, steps where someone has to read unstructured text or documents, and steps where a person waits for information that already exists somewhere. That is where AI automation pays back.
What can be automated
- Data entry from documents, emails and forms into your ERP and CRM
- Document verification and validation, with automatic checks and human review of doubtful cases
- Approvals and hand-offs between departments, with the required information already collected
- Periodic reports and dashboards generated automatically
- Certificates and conformity documents generated from test data
- Planning and cross-team coordination, with AI forecasts and suggestions
How we work
- MappingWe pick one painful procedure and time every step.
- A measurable goalWe define the number to improve in the process’s own units: quote turnaround, approval queue, hours of data entry.
- Built-in automationThe automation lives inside your existing systems, at the exact step where the work happens, not in a separate app.
- Human in the loopAI handles the easy cases and routes the hard ones to a person, with its reasoning attached.
- MeasureWe compare before-and-after numbers, and only then expand to other processes.
Real projects
My own AI agent platform: 23 templates for documents, phone and finance, delivered through a console, a public API and published web apps. Try it live: 3 free runs a day.
Enterprise .NET backend + applied AI: an AI document-verification system and internal-process automation that improve performance, scalability and reliability.
LLM assistant + RL scheduler embedded in a manufacturing ERP. Cuts cross-team coordination from hours to seconds.
CAN-bus reader & automated battery test certification, written for Voltfang's incoming-product test pipeline.
Technology
Further reading
Every company wants "automation". Few can say which process, in what order, and how they will know it worked. Here is how I decide what to automate — across manufacturing, finance and enterprise .NET backends.
Notes from a CRM modernisation in the Italian subsidised-finance sector. Why I keep recommending tiny AI features and bigger CRM investments.
Frequently asked questions
What is the difference between traditional automation and AI automation?
Traditional automation follows fixed rules on already-structured data. AI adds the ability to read documents, emails and free text, to classify and to forecast, so it can automate steps that used to need a person. In real projects the two work together.
Which process should we start with?
One your team finds painful and that has a measurable cost: usually a document-to-data step, or looking up information that today means interrupting a colleague.
Do we need to replace our business software?
No. Automation integrates with your existing ERP, CRM and backends through APIs and databases. I have worked on .NET, Angular, Python and SQL stacks.
How is the return measured?
In the process’s own units, not AI metrics: response times, hours saved, cases closed per day, errors avoided. The number is defined before starting and measured before and after.