Frequently asked questions
What clients ask us
most often
The questions we hear most often from clients, prospects and job applicants, with a short, honest answer. Can't find yours? Ask it below.
Getting started
How do we get started? Do we have to sign a big project right away?
No. We start with a free 30-minute call. After that comes a short discovery phase of three or four sessions: we learn how your business works, you get a list of around ten concrete improvements with a price range for each, and we turn the most important ones into a clickable demo. Only then do you decide whether, and with what, to move forward.
Can we test on a small scale before investing big?
Yes, and we recommend it. We first build a test or proof of concept on your own data, for example on a few hundred real emails. If it works, we build on it and none of that work is lost. If it doesn't, you know before the big investment.
Can you simply teach our team to work with AI, without a big project?
Yes. With an AI kickstart, a Sevendays expert works alongside your team for two days, at a fixed price of €2,450 excl. VAT. Your team learns to work with Claude and ChatGPT on your own tasks, and we set up a first agent or automation together. You decide how we spend the two days.
Can't you just give us a quick price?
A price without analysis is a guess: the same request can cost a thousand euros or seventy times that, depending on your systems and processes. After the discovery phase you get a well-founded price per component, so there are no surprises later.
Do you have experience in our sector?
We deliberately work across many sectors: construction, energy, transport, healthcare, manufacturing, insurance, cleaning, education, hotels and more. The processes are very similar: quotes, email, planning, reporting. What works in one sector, we bring to the next.
How long before it works?
Your first solution is live in 4 to 6 weeks. No six-month programme: after a few weeks your team is already using a first part, and we build on from there.
How much time does it take from our own team?
Count on two to three hours a week, mostly in the first weeks. Appoint one fixed contact person. We have a short weekly check-in and show each part as soon as it's ready, so you can adjust quickly.
Why not just use ChatGPT or Claude ourselves, or hire an in-house IT person?
That's a good start, but an empty chat doesn't connect to your mailbox, ERP or phone system by itself. That's exactly where the gains are. We're the team that builds and maintains those connections. If you have someone in-house working on AI, we're happy to work together: they know your business, we bring the technology and experience from other projects.
Pricing and payment
What is your hourly rate?
€105 per hour, or €840 per day, excl. VAT.
What does an AI solution cost?
That depends on what you want to automate and how many systems it needs to work with. You get a well-founded price after the discovery phase. We always weigh the cost against the return, for example the work you would otherwise hire an extra person for.
Do you work at a fixed price or on a time-and-materials basis?
Both. For a project you get a price per component up front, so you know where you stand. Small changes and questions afterwards are billed by the hour.
Do we have to pay everything at once?
No. For larger projects we invoice in instalments: at the start, at the first testable version and at delivery.
Are there monthly or hidden costs?
There are three kinds of costs, all on the table up front. You pay for the build once. Hosting and maintenance are a limited monthly amount, independent of the number of users. You pay for AI model usage directly to the provider (such as Anthropic or OpenAI), through your own account. We estimate it up front and keep an eye on it.
After delivery
What happens after delivery? Who helps us when there's a problem?
You keep one fixed point of contact you can call, email or message. In the first weeks after go-live we follow up closely. Small changes are billed by the hour. If you want something new, we make a separate proposal for it.
Do you offer an SLA?
By default we help as quickly as we can, usually within a few working days. If you need guaranteed response times, we put them in an SLA tailored to your business.
What if something doesn't work after delivery?
We fix bugs in what we delivered at no extra cost. Anything outside the agreed scope, we look at together.
Who owns the code? Do we?
Yes. The code lives in a private repository you always have full access to and can download in full. We build with open source, such as Laravel. Your data stays yours too: you can export all of it at any time.
Won't we become too dependent on you?
No. What we deliver keeps working without us, and another developer can build on it. You're not locked into one AI model either: switching to another model takes limited changes.
Your systems and what AI can do
Does it work with our software, such as Odoo, Teamleader, Exact Online or HubSpot?
Yes. We build on top of what you already have and connect through your software's API. Your software remains the source for customers, products and invoices. We work with Odoo, Teamleader, Exact Online, HubSpot, Microsoft 365 and Dynamics, Robaws and industry software, among others.
Do we need to replace our current system?
No. We connect, we don't replace. In fact, we advise keeping your software as standard as possible, so updates stay simple. We build the AI alongside it.
Our software has no API. Can it still work?
Usually, yes. We then start with a daily export to a separate data layer the AI works on. If an API becomes available later, we switch over.
Who is responsible when something goes wrong between all those systems?
We are. You have one point of contact. If a supplier changes its API, we update the connection and get in touch with that supplier ourselves.
We already have Copilot or ChatGPT. Isn't that enough?
They're useful for individual tasks, but you have to prompt them every time. They don't automatically handle every email, call or work order that comes in. We build AI into your workflow, so that work happens automatically.
Can AI read our mailbox and draft replies?
Yes. AI recognises the type of email (order, question, complaint), looks up the customer in your systems, routes the email to the right person and drafts a reply. By default nothing is sent automatically: a colleague reviews and sends.
Can AI enter orders from emails and PDFs into our ERP?
Yes. We first test it on your own historical emails, with your customer and product lists. What AI recognises with confidence goes through. Uncertain cases go to a colleague.
Should we build a customer portal?
Not always. Many customers keep emailing or calling anyway. Automatic status emails or a chat are often cheaper and work better. A portal does pay off when customers want to track quotes, work orders and invoices themselves.
Can AI read invoices, delivery notes or handwritten forms?
Yes. AI reads scanned and digital documents, including handwriting, and links them to the right order or customer. Anything it isn't sure about goes to a colleague.
Can AI help us create quotes faster?
Yes. AI can match a bill of quantities to your product catalogue and price list, take measurements from plans and draft a first version of the quote. A person always checks before the quote goes out.
Can AI answer the phone?
Yes. An AI phone assistant with a natural Flemish voice answers questions based on the information you provide, and can speak other languages too. We start with a trial to see how it works with your callers.
Can phone calls end up in our CRM automatically?
Yes, through your phone system or with a recording device such as a Plaud. The caller is recognised by their number, a summary lands on the customer record, and what you promised becomes a task.
Does it work through WhatsApp too?
Yes. The mobile number identifies the customer or employee, and processing works the same as for email. Customers can, for example, ask a question or make a booking via WhatsApp.
Can employees ask questions about our own company knowledge?
Yes. We bring email, conversations, documents and your systems together into one company brain you can query in plain language, with an answer and its source. Role-based permissions decide who sees what.
Can we get reports without the spreadsheet work?
Yes. We bring together data from your systems, so you can ask questions in plain language and get an overview of key figures and trends every morning.
Our data is messy. Does it need cleaning up first?
The better your data, the better AI works. But you don't have to clean everything up front: AI is good at finding duplicates and errors, and at getting new data entered correctly from the start.
How much time will it save me?
We measure that first: how much time it takes today. One example: checking supplier price lists used to take two to three days and now takes half a day to a day. Often the real question is: do you hire someone extra, or automate the work?
Is AI 100% accurate? And what about our less digital employees?
No, which is why we work with a confidence score: what AI is sure about goes through automatically, uncertain cases go to a person. For less digital colleagues we keep it as simple as possible, for example speaking a note after a visit instead of typing.
AI, data and privacy
Will AI be trained on our company data?
No. We only use business plans and APIs where the provider contractually does not train on your data. Free or personal accounts are a different story, which is why we advise against them for company data.
Where is our data stored?
Your data, such as your documents, search index and logs, is stored on servers in the EU by default. Only the piece of text needed to answer a question is sent to the AI model.
Can the US government access our data?
The strongest AI models come from US companies, and the CLOUD Act lets US authorities request data from US providers in certain cases, even when it's stored in Europe. That's why we limit what is sent to the model and keep your data itself in the EU. If you want to work fully European, that's possible with European models, though they're less capable today.
Can we run AI locally or on our own server?
Yes, with an open-weights model. But it requires heavy hardware and the quality is lower than the large models. Local feels safer, but the big cloud providers often secure their systems better than a server of your own.
Do you sign a data processing agreement?
Yes. The data processing agreement is a standard annex to our contract, together with the list of external providers (sub-processors) we use.
Can I record conversations? Do I have to say so?
We always ask for permission up front, and don't record if someone doesn't want us to. What the GDPR and Belgian law require exactly is set out in our knowledge base article.
Read more: Recording conversations: what's allowed in Belgium?
Can all phone calls be recorded and summarised automatically?
Technically yes, through your phone system. We only process calls with contacts from an open quote or project, and filter out private calls. What you need to disclose is covered in our knowledge base article.
Read more: Recording conversations: what's allowed in Belgium?
Can AI read employees' mailboxes?
We start with shared mailboxes, such as info@ or service@. Personal mailboxes are only added with the employee's agreement, and with clear rules on what is and isn't processed.
What does the AI Act mean for us?
If you use AI as a tool, for email, reports or searching your documents, you fall into the lightest category. Your team does need sufficient AI literacy, and people must know when they're talking to a chatbot or phone assistant. What applies when is covered in our knowledge base article.
Who sees what? Can I set permissions per person?
Yes. AI inherits the user's permissions: anyone who can't open a document today won't see it through AI either. Sensitive information such as salaries or margins is restricted per role.
Can AI send emails or make payments on its own?
Not without a person checking. AI doesn't pay anything or send anything out automatically without review. Sensitive actions always require confirmation.
Can employees just use free ChatGPT?
Not with company data. With free accounts, your input may be used to train models. Choose one business setup for your whole team, and keep personal and work accounts separate.
AI models and costs
ChatGPT, Claude, Copilot or Gemini: which should we choose?
For business work we currently find Claude the strongest, but the differences are small and shift quickly. That's why we choose the right model per task and build so you can switch later.
We already have Microsoft Copilot. Do we need to drop it?
No. Copilot remains useful in Word, Outlook and Teams. What we build works alongside it. You can now also choose Claude inside Copilot.
What if a better model comes out tomorrow, or the price doubles?
Your data and logic aren't tied to one model, so switching stays possible. We don't switch on a whim: we review it with you first. And take monthly subscriptions, not yearly ones.
Subscription or API: what's the difference?
You pay for a subscription per user per month, for people working in a chat themselves. With the API you pay per use, and that's what you need for anything running automatically in the background, such as email processing.
How much does AI usage cost per month?
That depends on how much is processed. Usage runs through your own account with the provider. We estimate it up front, build in a limit and give you a monthly overview, so you're never caught by surprise.
Which Claude plan should our business choose?
For a team we recommend Claude Team: an admin manages settings and shared instructions for everyone, and your data isn't used for training by default. Take it monthly.
Do I need a Plaud subscription?
Not if we process your recordings: we fetch the audio and turn it into a report and tasks ourselves. If you want to use Plaud's own processing, you do need their subscription.
Working at Sevendays
What does the application process look like?
You apply through the form or by email. Then comes a short call or video call. After that you visit our office for a session of one to one and a half hours: you dig into a real project and discuss it with a colleague. There's no coding test, and you're welcome to use AI. We mainly look at how you reason.
Do you hire juniors or recent graduates?
Yes. How you work and think counts for more than years of experience. Whether there's room for a junior at any given moment does depend on workload, because good mentoring takes the team's time.
What technology do you use? Do I need to know Laravel?
Laravel is our foundation, with Filament, Livewire, Alpine and Tailwind. The work is strongly focused on backend, integrations and AI. If you don't know Laravel yet, you'll learn it here.
Am I allowed to work with AI?
Yes, it's the standard. Everyone at Sevendays works with Claude Code every day, including people who don't write code.
Will I have direct contact with clients?
Yes, a lot. We don't have separate project managers: as a developer you take over a project after the sale, from analysis to delivery, and you talk to the client yourself.
Can I work from home?
Yes, there's no obligation. In practice most people come to the office about three days a week. Every morning at 9:30 we have a short stand-up, which people working from home join too.
What does a working day look like?
The day starts with the stand-up at 9:30. On Mondays we divide the work for the week together. You work through tickets on your own projects, and every few weeks there's a team meeting.
Can I do an internship with you?
We don't currently offer internships.
Can I apply from abroad or without speaking Dutch?
You need to speak Dutch fluently and live in Belgium. You'll work with Flemish clients every day, and we see each other regularly at the office in Antwerp.
About Sevendays
How big is the team?
We're a team of seven, a permanent group of local experts in Antwerp, active since 2016. No subcontractors and no offshore: the people who think along with you also build.
Where is your office, and how do I get there?
Regine Beerplein 1 E206, 2018 Antwerp, about a ten-minute walk from Antwerp-Berchem station. Our directions page shows step by step how to get in.
Can't find
your question?
Rather talk right away? Book a call · Call · Email