A.I. that makes your agency faster.

No. The question isn't how big you are, but how many times a week the same work comes back around.

If you're three people closing two sale agreements a month, a document generator isn't a priority. If you get calls every single day about a leak, a key or an invoice, that is a priority, even at your size.

That's why we always start with a session where we lay your processes side by side and put numbers on them. So you know up front what pays off and what doesn't.

That depends on your software. We'd rather tell you this beforehand than afterwards.

With Keypoint we have a working integration running at a property management firm. Omnicasa is fairly open and allows integrations easily. With Zabun and Whise it depends on the API and on the arrangements with the vendor. At Zabun we've had to knock on support's door ourselves in the past to get API access sorted. For Rentio, PlanRadar, 3CX, Microsoft 365 and SharePoint we work with existing integrations.

If we can't get an API, there's usually still a way: an export, a shared folder or a form as an intermediate step. But then it's slower, and we'll say so.

Send us the name of your software and you'll have an honest answer within the day.

A.I. can produce a draft. You're the one who signs.

Here's how it works: your template stays your template. The agent reads the documents already in the file, such as the identity details, the energy performance certificate, the soil certificate and the cadastral info, and fills the fields from those. A sale agreement easily runs to 140 fields across 14 sections. Anything the agent isn't sure about, it flags visibly, so you know where to look.

One agency we did the maths with put the saving at roughly an hour per agreement themselves. That's their figure, not ours, but it gives you the order of magnitude.

What A.I. doesn't do: decide what a clause should say, or judge whether a special condition holds up legally. That stays human work.

Not straight away, and usually not entirely.

We almost always start behind the phone rather than in front of it. A human still takes the call, but after that it continues automatically: the caller is recognised by their number, the call is transcribed, a short report with proposed tasks comes out, and the property manager reviews it before it goes into the file.

At Immo Cauwe, a property management firm in Knokke, around 1,600 calls a month come in this way. Those reports and tasks now land automatically in Keypoint and SharePoint, after the manager's review.

Only after that do we look at an A.I. phone agent taking the first line. And there too: the caller decides. If someone says they want a human, the bot puts them straight through.

Yes, but not just like that. Two things get tangled up here.

If you're taking part in the call yourself, you may record it in Belgium for your own use. The moment you keep, transcribe and process that recording, the GDPR applies. Then you have to inform your caller, you have to be able to explain why you're doing it, and you have to set how long you keep it.

In practice we solve that with a short notice at the start of the call, a fixed retention period and a data processing agreement. We aren't lawyers, so for the final legal check we point you to your own counsel or your professional federation.

No, and we don't just do that.

The portals' terms of use generally don't allow you to read their listings automatically. Technically it's possible; legally that's another story. We'd rather be honest about it than leave you stuck with it later.

What is possible: working with official APIs and data partners, working with your own CRM data, or using a public source that does allow it. If you want market insight, we'll look together at which source you may legally use and what it costs you.

Sorting yes, selecting no. And that's not us being cautious, that's the law.

In Belgium you may not turn down prospective tenants on the basis of protected characteristics such as origin, age, family composition, disability or the fact that someone is on a replacement income. An A.I. that ranks people from an email can pull those characteristics in without anyone noticing. We don't take that risk.

What we do build: an agent that collects every response, strips out duplicates, requests missing information and shows per candidate which documents are in. Everyone gets an automatic reply, including the people who don't get it. The decision itself stays with you, and afterwards you can show which criteria you made it on.

A real valuation? Not with a chatbot alone.

During a training session we watched an agent literally ask: give me a valuation for this property. What comes out looks convincing and is worth nothing. The model has no reference points in your street.

What does work is a valuation tool that runs on real data: median prices in the area, your own sales history, the property's characteristics. Agencies mostly use those on their website today, to give visitors a first indication and pull in listing leads that way. It stays an indication, not a site visit.

You, and that doesn't change. An A.I. agent is a work tool, not an employee with liability of its own.

That's why we don't build towards a system you have to trust, but towards a system you can see into. You get to see where a piece of data came from. The checkpoint sits where the risk sits, not everywhere. And anything that goes straight to a client, an owner or a notary, we first test in silent mode: the agent does its work, but you only see what it would have sent. Only once you say it's right do we switch sending on.

You stay the owner. We're only the party working with it on your instructions.

We sign a data processing agreement at the start of every engagement, not afterwards. It sets out who can reach your data, where it sits, how long we keep it and what happens if you stop. Processing stays within Europe and your data is never used to train A.I. models.

For sensitive files we don't work with public chat tools. That's an arrangement we make up front, not something you find out about later.

For images and video: yes. Since 2 August 2026, the European A.I. Act requires you to make content that A.I. generated or altered recognisable as such. That can be done with a note or through the file's metadata.

For a text you review and publish yourself, it's less strict. But our advice is simple: put in your house rules what is and isn't allowed, and make sure every employee knows when a label is needed. That's exactly the kind of arrangement we set down in an A.I. governance programme.

Not because it's A.I., but yes if it's bad. Publishing mediocre text at scale works against you, with or without A.I.

What does work, we see at an agency in Spanish property. There, an agent writes articles of 1,200 to 1,500 words on keywords that come out of Google Analytics and Search Console. A second agent optimises them for search engines and for language models such as ChatGPT and Claude. And then the important part: the copywriter gets it as a draft, reviews it, adds their own knowledge and only then publishes.

The goal isn't ten times more articles. The goal is one good article every two weeks, without hiring someone for it.

Absolutely. Our most developed property case is a property management firm.

At a property manager the pressure sits on the phone and on the follow-up. An owner calls about a leak, a tenant about the hallway, a supplier about an invoice. Every call then has to end up in the right building and the right unit. That's exactly what an integration between your phone system, your meeting notes tool and your property management software solves.

On top of that we see a lot to gain at property managers in preparing the general assembly, in following up contractor quotes, and in the questions that keep coming back and that a knowledge base can answer perfectly well.

Less than you think, but not zero.

What takes time is the start: half a day together to map out your process, and after that a few short sessions to get the settings right and test on your own files. Count on a fixed contact person on your side who gets time for it.

What takes little time is the using. If it's built well, what mainly changes for your managers is what they no longer have to do.

No. Anyone who can work with property software can work with this.

We deliberately build so nobody has to learn prompting. You fill in a form or click a button, and you get back a draft you review. That's a deliberate choice, because almost every agency has two kinds of people: the enthusiast who wants to try everything, and the colleague who says they want to do their job and not get involved with IT. It has to work for that second person, otherwise it doesn't work.

We see that a lot. An agency rolls out ChatGPT or Copilot, and a few months later half the team has stopped using it. Usually that isn't the tool, it's two things: nobody owns it, and too much arrived at once.

What does work: one internal owner, one process at a time, and a fixed moment where we look together at what's sticking. And a clear framework from the top. Agencies that say "this is how we work from now on" get better adoption than agencies that leave it optional.

The applications we build most in the sector today:

• Turning phone calls automatically into a report, tasks and the right file in your CRM.
• Triage on incoming calls, so that what can wait really does wait until the next working day.
• Drafting sale agreements and rental contracts from the documents already in the file.
• Requesting and chasing documents from the owner, the notary, the photographer and the council, with automatic reminders.
• Reading identity cards and certificates automatically instead of retyping them.
• Social media: automatically queueing up a post the moment a new property goes online.
• Writing blogs and e-books based on your own keywords and house style.
• A quarterly report on your competitors: their new content, their social channels and their Google reviews in one document.
• A valuation bot on your website to generate listing leads.
• Sorting the general mailbox and drafting replies to the questions that keep coming back.
• A dashboard that puts your ad budget and your website visitors next to your leads.
• Assembling the monthly feedback to your sellers automatically from the viewing notes.

We prefer to start with an A.I. Opportunity Mapping. In it we map every opportunity together with your team and plot them on value against complexity. After that we work one idea through completely in a business case. So you know up front what it returns and in what order you're best off starting.

After that we work in phases, with a decision point after each one. You start small, with one branch or one process for instance. If it works, we carry on. If it doesn't, it stops there and you haven't paid for the full programme.

For a first built flow you'll usually see a working version within four to six weeks. We only take on projects we believe will return more than they cost in the first year.

Yes. Not every agency needs to have something built straight away.

Plenty of agencies start with an inspiration session for the whole team, to show what's possible in the sector today, or with training where your people learn to work with ChatGPT, Copilot or Claude on their own files. Our standard meeting notes tool is Leexi.

Our advice is to flip the order, though. First do a short session with your management to know what's feasible, and only then the session for the team. That way you don't promise your people things that technically can't be done because your software has no API.

Yes. We're not an extension of any package and we get no commission for pushing you somewhere. We work inside what you already have.

That also means we'll sometimes advise against having us build something, because your own software solves it better or because there's an existing tool that's cheaper. You'd rather hear that from us than from someone with a stake in it.

There is one. A.I. doesn't free up time out of thin air. The gain sits in the preparatory work, and that gain only becomes money once you also decide what you do with the time it frees up.

And the integrations are often tougher than a demo suggests. At one property management firm, 35 to 40 percent of calls turned out not to be automatically matchable, simply because phone numbers were sitting in the notes field instead of the phone field. That kind of work sits under the bonnet and doesn't show up on a slide.

That's why we measure up front how much time a process really costs today, instead of working from assumptions. If the sums don't add up, we'd rather say so now than after three months.

There are tools that do one thing for property agencies, such as a chatbot or a valuation tool. If that works for you, by all means take it. We're not building a second version of it.

Where we make the difference is in what runs across your packages. Your telephony, your CRM, your SharePoint, your Outlook and your marketing aren't connected today. That's the work that gets left, and that's exactly what we do.

We work today for firms in estate agency, property management, property stewardship, project development and property abroad. We also run A.I. training for Immoversity, the academy for estate agents.

Want hard figures about our company? Ask in the conversation. We'll put our balance sheet on the table.