AI · TASK AUTOMATION
AI automation. The tasks you repeat every day, made easier.
Sorting emails, preparing quotes, writing meeting notes, pulling the details from an invoice: I set up AI automations for this repetitive work. You keep the final say on anything that commits your business.
01 / THE TASKS Less copy and paste, more useful time.
AI can read a text, sort it, summarise it and pick out the information it contains. Connected to your tools, it takes over the mechanical steps of work you currently do by hand.
- Email sorting. Incoming messages are read and filed by type: quote request, question, invoice, complaint. For routine cases, a draft reply is waiting for you to check before it goes out.
- Quote preparation. From a request received by email or through a form, the useful details are picked out and a draft quote is prepared with your prices. You check it, adjust it and send it.
- Meeting notes. Your notes or a meeting transcript become a clear summary: the decisions made, the actions to take and who is in charge of each one.
- Document data. Invoices, purchase orders, PDF forms: dates, amounts and references are extracted and filed in a spreadsheet or, where possible, in your software, instead of being typed in again.
- Regular reports. Figures from several sources are gathered and presented in a report written in plain language, every week or every month, ready for you to review.
Another task? A task that comes up often, follows fairly stable rules and gives a result that is easy to check is a good place to start. Describe it to me and we will see together whether it is a good fit.
02 / THE METHOD Check first, then extend.
I start small, on a concrete case, and only widen the scope once the result has been checked with you.
- Look at the task. You show me how it is done today, with which tools and which exceptions. Together, we decide what will be automated and what stays in your hands.
- Test on your examples. A first trial runs on real cases from your business. You see the results, mistakes included, before any decision is made.
- Go live. The automation is connected to your tools, with human approval wherever a mistake would be costly. Simple documentation explains how it works.
- Monitor and adjust. At first, the results are reviewed closely. The rules are corrected where needed, then we can move on to the next task.
03 / THE LIMITS What AI won’t decide for you.
AI is useful for repetitive work, but it is not infallible. I would rather explain its limits before we start, so the automation is built around them.
- It can make mistakes. A misfiled message, a misread figure, an invented detail: these errors happen. Steps that commit you, such as sending a quote, always go through your review.
- It does not replace your judgement. A sensitive complaint or a commercial decision needs your eye. AI prepares, you decide.
- Your data calls for care. I explain where your information goes and keep what is sent to the minimum needed. The choice of tools takes the sensitivity of your data into account.
- Tools change. AI services change their features and prices. I favour simple, documented setups that can be modified or switched off without rebuilding everything.
A well-designed automation saves you time on the mechanical steps and leaves the decisions to you. I will also tell you when a task is not worth automating: too rare, too variable or too delicate.
04 / WHAT IS INCLUDED The points I keep under control.
- A precise scope and a schedule before we start
- A trial on your real cases, mistakes included
- Human approval wherever a mistake would be costly
- Data shared only as far as strictly needed
- Your tools and accounts in your name
- Simple documentation of how it works
Your questions.
Which tasks can be automated with AI?
Mainly recurring tasks that follow fairly stable rules and give a result that is easy to check: sorting emails, preparing draft quotes, summarising meetings, extracting information from documents or producing regular reports.
Can AI make mistakes?
Yes. It can misfile a message, misread a figure or invent a detail. That is why I test each automation on your examples before it goes live, and I plan a human review at the steps that commit your business.
What happens to my data?
We choose the tools together, based on how sensitive your information is. I explain where it goes and keep what is sent to the strict minimum. As a cybersecurity engineer by training, I deal with this question from the start.
Do I need to change my tools?
Not necessarily. I start from what you already use, such as your email, your spreadsheet or your invoicing software, and connect them where possible. If a tool does not allow it, I tell you and we look for another solution.
What is the difference between an automation and an AI agent?
An AI agent holds a conversation: it understands a freely written request, then answers or hands over to you. An automation chains steps defined in advance between your tools, often without any conversation; AI is used there to read, sort or write. For a repetitive, well-defined task, an automation is often the simplest solution.
How long does it take to set up an automation?
It depends on the task and on your tools. After a first conversation, I propose a precise scope and a schedule before we start.
How does a first conversation work?
You write to me on WhatsApp or by email, describing the task you would like to take off your plate. I share a first opinion, then we look together at how the task is done today.
YOUR PROJECT · ONE POINT OF CONTACT
Is a task taking up too much of your time? Let’s talk.
Tell me about this task and the tools you use. I reply with an initial assessment and the next steps.