Service · Artificial intelligence
AI applied to your business, not to the hype
Automation handles what is a rule. AI handles what requires judgment: reading, classifying, summarizing, deciding between similar cases. I help you pick the right task and build the solution fitted to the tools your company already uses.
Applied AI means using language models to run tasks that require interpretation: triaging messages, classifying tickets by their text, summarizing documents, extracting data from PDFs. Unlike automation, which follows a fixed rule, AI deals with what varies case by case. I diagnose which task is worth it, build a small proof, and only then scale.
How do I use artificial intelligence at my company?
By starting with a single task. The one that works best has three traits: it is repetitive, it requires some interpretation (it is not a fixed rule), and mistakes are cheap to correct because someone reviews the output before it becomes a decision. Triaging incoming messages, classifying a ticket by its text, summarizing a long contract, pulling data out of PDF invoices: all of those fit the description.
What does not work is the opposite path: subscribing to a generic AI tool and then hunting for somewhere to use it. It becomes one more subscription nobody opens. I do the reverse. I look at your process, find the judgment task that eats the most time, and build on top of that. If I cannot find one that pays off, I say so. Saving you the investment is worth more than selling you a project.
What can AI actually do today?
Triage and classification
A message, ticket, résumé or order arrives and has to be read before anyone knows where it goes. AI reads it, classifies it and routes it.
Long-document summaries
Contract, report, minutes, assessment: the key point and the risks in a few lines, quoting the original passage.
Data extraction
Invoices, PDFs, emails or forms turned into structured fields in your system, with nobody retyping anything.
An agent with your company context
An assistant that answers the way your operation works, because it received your instructions and your documents, not the generic version from the internet.
Memory across conversations
Context systems so the AI does not start from zero every time, keeping what has already been decided.
Assisted support
A draft reply ready for a person to review and send. You gain speed without taking the human out of the decision.
How do we get started with AI?
Assessment
We talk about where your team spends time reading, checking and deciding case by case. I tell you what AI solves, what plain automation solves better, and what is not worth touching.
A small proof
We pick one task and I build the minimal version with your real data. You measure the result before investing in the rest.
Fitted to what exists
Did it work? I connect the solution to what you already use (spreadsheet, ERP, email, WhatsApp, web systems) and keep a record of what the AI did in each case.
Scale with control
Then we expand: more volume, more tasks, more people using it. Always with human review wherever a mistake is expensive.
Not just theory
I build with AI every day and publish the result. You can open, use and test these right now, no signup:
What is the difference between automation and artificial intelligence?
Automation handles rules: if the order has an invoice, mark it as billed. The condition is always the same and the software just executes. AI handles judgment: reading 200 résumés and separating the ones that make sense, telling from the text whether a ticket is urgent, summarizing a 40-page contract. No fixed rule covers every case, so it has to be interpreted. In practice the two work together: AI interprets and automation executes what comes next.
Frequently asked questions
How do I use artificial intelligence at my company?
By picking a single task to start with: repetitive, requiring interpretation, and where a mistake is cheap to correct. Message triage, ticket classification, document summaries and PDF data extraction are the cases that pay off most often. Based on how that first one performs, we decide whether to expand.
What is the difference between automation and artificial intelligence?
Automation executes a rule: the condition is always the same and the software just follows it. AI deals with judgment, like reading a text and deciding where it goes, summarizing, or comparing similar cases. In practice the two combine: AI interprets and automation executes the next step.
Is AI useful for a small company?
It is, and sometimes that is where it pays off most. The gain scales with how much time someone spends reading and deciding case by case, not with company size. If one person spends hours a week triaging messages or checking documents, the return shows up fast.
Will the AI see my customers' data?
Only what is necessary, and that is a design decision, not a technical detail. Data can be anonymized before it is sent, only the relevant field can be processed, or a model that does not train on your data can be used. Privacy law requires you to know which data leaves, where it goes and how long it is kept, so I put that in writing before anything is switched on.
Do I need to replace the systems I already use?
No. AI sits at the edge of what already exists: it reads from your email, your spreadsheet or your system and returns the result in the same place. Replacing a system is a separate project, and it is rarely what solves the problem that brought you here.
What does it cost and how does it start?
It starts with a free assessment conversation where I tell you what is feasible and what is not worth doing. Then comes a small proof with fixed scope and price, so you can measure the result before investing in the rest. Price varies with volume, how much integration is needed, and how much human review the process requires.
Keep reading
Which task at your company requires reading and judgment?
Tell me about the process where someone has to read, interpret and decide case by case. The assessment is free and, during the conversation, you will already know whether AI solves it, whether plain automation solves it better, or whether it is not worth it.