AI in your business: what it already does today

AI automation means software takes over work that used to require someone to read or listen: an email, a scanned ticket, a floor plan or a dictated report. Aziri builds that AI into the custom applications it develops for businesses. AI proposes, a person confirms.

Some of our clients

  • Spaas
  • Itzu
  • Warsco
  • DCM (De Ceuster Meststoffen)
  • Viabuild
  • Van Loo Projects
  • Eddie Stobart
  • Boels Verhuur
  • Moore
  • Sweco
  • Zwembaden Bollen
  • ValoDirect
  • Varey
  • Hamster Cleaning
  • A-S-W
  • Alracon

What does AI do in a company?

In a company, AI mostly takes over retyping and looking things up: work orders typed up in the evening, orders from an email that have to go into the ERP, measurements from a drawing that end up in Excel. AI has recently become able to read things that have no fixed form, and that is where the gain is.

Aziri does not sell AI, we solve a problem. In everything we build, it works in the same four steps.

  1. Something comes inAn email, scan, dictated report or drawing.
  2. AI makes a proposalFields, a draft or a priority, with its confidence.
  3. A person confirmsDoubt shows in red. Nothing leaves without approval.
  4. It is in your systemIn your ERP, schedule or file, without retyping.

What is the difference between AI automation and regular automation?

Regular automation follows fixed rules; AI automation can read and understand what has no fixed form. Often they sit in the same application.

TaskRegular automation or integrationAI automation
Processing an orderWorks if the customer orders through a form or EDI.Reads an order from a free-form email or PDF and creates a draft order.
Booking an invoicePeppol invoices come in structured and need no AI.Useful for foreign invoices and documents that are not Peppol.
Filling in a work orderA digital form with drop-down lists.The technician dictates what he did; AI pre-fills the fields.
Answering a questionA fixed FAQ or a keyword search.An answer from your own reports and procedures, with the source included.
Making a quoteA calculation with fixed formulas and price lists.Reads the request or the drawing and prepares a draft with your own items.

If a form or an integration solves it, that is cheaper and more predictable. And we will tell you so.

What does AI look like in your software?

In your software, AI shows up as fields that are already filled in. Your people check, adjust where needed and confirm.

Illustration: the technician dictates, AI pre-fills the work order.
Illustration: each field gets a confidence score; doubt goes to a person.
Illustration: the answer comes from your own documents, with the source.

34 uses across seven industries. Pick your industry and open a use to see what it does.

AI in construction

For contractors, it is mostly about what is said, drawn or handed over on site and retyped at the office afterwards. More on AI in construction and on our construction page.

Dictating a site visit

After a site visit, the owner dictates what he saw. The app transcribes it, summarises it per topic with action points and files it on the job. Whoever calls later finds the report on the file, not in someone’s head.

More on AI in construction

An assistant that asks what is missing

The owner first talks freely, then an AI assistant asks for what is still missing and fills in the site visit form. It only saves after approval. That way nobody forgets the crane access or where the electricity box is.

Checking building rules per plot

You enter an address. The tool fetches the plans and regulations that apply to that plot, scanned ones included, and checks them against your own checklist with a quote and page number. An employee confirms. You know before the first sketch whether a project fits that plot.

From specification to bill of quantities

AI pulls items, quantities and units from a specification and prepares them in your estimate. When it is unsure, it flags the item so the estimator knows where to look.

More on AI in construction

Reading earthmoving delivery tickets

A photo of the ticket is enough: trip, haulier and m³ are linked to the job straight away, without retyping.

Subcontractor invoice next to site hours

AI compares the invoiced hours with what the crew registered on site and flags a difference, so you can ask before you pay.

Spotting extra work

If the work order or site report mentions something that was not in the quote, the app asks whether it is extra work. You agree it in time and invoice it.

AI for installers and field service

For an installer, the work is in reading drawings, closing work orders and sending the right technician on the right day. See also the digital work order and our planning software.

Dictating the work order, with translation

The technician dictates what he did and the text lands in "Work carried out" on the work order. If he speaks English, French or Czech, a Dutch translation appears below. Whatever he typed himself stays.

More on Digital work orders

Reading a floor plan for the quote

On a floor plan, AI recognises each room with its name, area and perimeter, even on a scanned drawing. The estimator gets a list per room instead of a drawing to measure, and confirms or adjusts.

More on Reading documents with AI

Estimating how urgent a request is

AI reads what the customer wrote, such as "water in the basement" or "pump stopped in the frost", and proposes a priority with the reason. A priority set by the planner always wins.

More on Planning software

Maintenance and repairs scheduled automatically

The app assigns each new work order to the best-suited technician and reshuffles every night. Urgent jobs move forward, and technicians criss-cross less. This is a scheduling algorithm, not AI.

More on Planning software

Quotes and emails in the customer’s language

Quotes, work order PDFs, emails and text messages go out in the customer’s language automatically. Product names and codes stay untranslated.

More on Quotes with AI

From electrical drawing to material list

From an electrical drawing, a list of sockets, switches and light points per room as the basis for your quote. With low confidence, counting yourself is just as fast, so we test on your own drawings first.

AI in transport and logistics

In transport, much arrives by email and on paper: bookings, CMRs, weighbridge tickets. See also our transport planning software and the transport and logistics industry.

From booking emails to a transport schedule

Customers book trucks by email. For each email, AI proposes the transport lines: customer, article, date, haulier, number plate and weight. An employee confirms, and the lines land in the schedule instead of Excel.

More on Reading documents with AI

Reading CMRs and weighbridge tickets

From a single PDF of dozens of pages, AI separates the CMRs and weighbridge tickets and extracts number plate, weight and date, with a confidence score per field.

More on Reading documents with AI

Rescheduling when something drops out

If a driver or truck drops out, the app proposes which trips go to whom. The planner decides and the drivers see it on their phone straight away.

Answering "where is my delivery?"

An assistant answers the customer from your own schedule and trip data, and hands over to a colleague as soon as it is about a price or a complaint.

AI in wholesale and distribution

In wholesale, most manual work is retyping orders and requests. See also wholesale and distribution.

Orders from email and PDF into your ERP

An order in an email, an Excel file or a PDF becomes a draft order with your article numbers. An employee checks the doubtful cases and sends it on to your ERP.

More on Reading documents with AI

Summarising a quote request and proposing prices

AI summarises a request, finds the matching articles in your own catalogue and prepares the prices from your price list. The salesperson adjusts and sends.

More on Quotes with AI

Seeing a stock shortage coming

From your sales history and open orders, the app predicts which articles will run out in the coming weeks, as a proposal for the buyer. See also inventory management.

AI in manufacturing

In manufacturing, a lot of time goes into reading orders and drawings, and into knowledge that sits in one person’s head. See also manufacturing.

Reading customer orders and drawings

An order or technical drawing in PDF becomes a draft work order with dimensions, material and quantities, ready for production planning.

Asking for work instructions on the floor

An operator asks on a tablet how a setting should be done and gets the answer from your own instructions and reports, with the source.

More on Company knowledge with AI

Predicting maintenance from machine data

Vibration, temperature or running hours show that a part is wearing, and the app schedules maintenance before the machine stops. The machine does need to record that data already.

AI in rental

In rental, it is all about what goes out, what comes back and in what state. See also rental.

Spotting damage on return

Photos at departure and return are put side by side, and the app marks what looks new. The employee decides what to charge.

Rental requests from email into your schedule

A request by email becomes a draft booking with equipment, dates and address, and the app shows straight away what is available.

Tracking inspection certificates

Inspection reports for aerial platforms or cranes are read, so the expiry date is in the app and no machine goes out without a valid inspection.

AI in the office: admin, HR and sales

These uses fit every industry: wherever emails, reports and questions pile up.

Answers from your own company knowledge

Processes, agreements and reports live in one place, and an assistant answers with the source. A new colleague finds out how something is done without interrupting the one person who knows.

More on Company knowledge with AI

Customer emails on the right file

A forwarded customer email lands with the right customer, project and ticket, or becomes a new ticket. Nobody has to work out which project an email is about.

Minutes and tasks after a meeting

After a customer meeting, the app writes the minutes. Action points become tasks in one click, after checking they do not exist yet. It works the same for a site meeting.

Quote text and follow-up email from the customer file

AI writes a first quote text from what the customer said and emailed in recent months. The description is much faster that way; the salesperson always checks the price and hour estimate.

More on Quotes with AI

Preparing prospecting

The app looks a company up and builds a research card and a first email draft. A check blocks unproven claims, and a person sends.

Reading a safety certificate or licence

From a photo of the certificate, AI fills in the expiry date in the staff file. HR and planning see in time who needs a new one.

Explaining pay rules in plain language

"Why did I get this allowance?" An assistant explains it from your own rules and the registered work, so HR does not answer the same question ten times.

Pre-sorting job applications

AI summarises CVs and sets them against your vacancy. A person makes the decision, and candidates know AI is involved.

Where do you start with AI, and when do you not need it?

Start with one task that comes back often

  • It costs hours every week.
  • It is roughly the same each time.
  • Someone can check the result.

Retyping from emails, tickets or drawings is usually the best start. Measure in advance how long it takes now, so you know afterwards what it delivered.

No AI if …

  • a form or integration will do. If a document always comes in the same form, a regular app is cheaper and more predictable.
  • the process is not yet settled. AI speeds up a process, but it does not resolve confusion.
  • the volume is too small. Retyping ten tickets a month costs less than building something.

Is AI safe for your company data?

Your data is stored in your own application. For processing, only what is needed goes to a model provider’s business service, which does not use it to train models. Nothing goes to a customer or the books without an employee’s approval.

What happens to our data?

For the AI we use the business services of model providers such as Anthropic (Claude), OpenAI, Google (Gemini) and Mistral. By default, they do not use business data to train their models. If processing must take place in the EU, that is possible through a European region. Which model and which region we use is set out in the analysis.

What if the AI makes something up?

It can happen, so we build for it. The AI shows where it is in doubt and would rather leave a field empty than make something up. It does not overwrite what someone filled in themselves. We first test every use on real documents or recordings from your business.

Does it work with our ERP or accounting?

Yes, if it allows an integration. We have connected to 70 systems so far, such as Exact Online, Odoo, SAP and AFAS. See integrations.

Who owns it?

The code we write for you and all your data are yours. If part of it runs on a third-party service, such as a language model, that is stated in the proposal. We build so that you can switch models.

How does Aziri itself work with AI?

We use AI every day, when building software and in our own platform Hippo. A project lead checks every result like the work of a colleague. We gather knowledge of your process at the table with you, not from a model.

  • What used to be months of programming, we now build in weeks, with tests and a review.
  • Hippo writes up reports after meetings and puts customer emails on the right ticket.
  • A knowledge base brings together all our agreements and processes. See company knowledge with AI.
  • AI agents prepare research and first drafts. Nothing goes online without a person’s approval, this page included.
Two Aziri colleagues looking at a result on the laptop together

Frequently asked questions

Question not listed? Ask us directly.

What is AI automation?

AI automation is software that takes over tasks where someone had to read, listen or decide based on text: linking an email to a file, reading a ticket, writing up a report. Classic automation follows fixed rules; AI can also work with varying formats and free text.

What does AI cost for a business?

From €2,000, depending on what has to be built around the AI: the screen to review proposals, the integrations and how many kinds of documents or emails are involved. The usage cost of AI services is small and we estimate it in advance, based on your volumes. Preferably you get one fixed price for the build.

Which AI uses are there for businesses?

The most common AI uses are reading documents (tickets, order emails, drawings), dictating a work order or report, preparing quotes, sorting emails and linking them to the right file, and a knowledge base that answers questions from your own data. On this page they are listed by industry.

Can’t we just use ChatGPT or Copilot?

For writing, summarising or translating, often yes. But ChatGPT does not know your customers, prices and agreements, does not work in your ERP and has no rights per user. As soon as AI has to fill something in in your own software, custom work is needed. ChatGPT Business, Copilot or custom AI?

Do we need to do anything for the AI Act?

There is European AI legislation, the AI Act. For the uses on this page, it mainly means making clear that AI is involved and letting a person decide. We take it into account when we build. More in AI Act for businesses.

Is there a subsidy for AI?

In Flanders, a truly innovative project can qualify for a development project from VLAIO. Sirris and Agoria organised coaching for SMEs with Start AI; check whether there is a new round. In Wallonia, Start IA and Tremplin IA ran in 2026, but those calls are closed; follow Digital Wallonia for the next one.

How quickly is a first use up and running?

A first working version is usually ready within 2 to 6 weeks, as part of the application. We first test on real examples from your business, because only then do you see how often the AI is right.

Where does someone in your business type or search too much?

Tell us which work gets stuck today. We will tell you honestly whether AI adds anything there, or whether a regular app or integration will do.

We reply quickly.