About

Who is behind Harbour

Harbour is made by At Ease AI Technologies Pty Ltd, a Melbourne company with one founder, who writes the software.

He came from physics. Then consumer software, where the test is whether a stranger can use it without being taught. Then commercial software, where the test is whether it survives somebody else's rules.

Before Harbour he built an energy platform. It assessed commercial and industrial solar, batteries and power purchase agreements from real electricity data. That meant reading interval meter data, parsing electricity bills, and handling retailer letters of authority. It produced return-on-investment reports. That platform is still running. It is no longer being developed.

That work is why Harbour is built the way it is. A misplaced decimal in a return-on-investment report is not fixed by an apology afterwards. The same is true of a phone line in a licensed industry.

At Ease Properties Pty Ltd is a licensed Victorian real estate agency in Clayton, ABN 83 691 008 714. It is a shareholder in the company, and it is Harbour's first customer. The people there who answer the phone, in sales and in property management, will use Harbour before any other customer does, and the agency they work for owns part of the company that builds it.

What we are building

We are not going to tell you we have rethought real estate with AI. It is too early for that word, and we have not earned it. We have one product: an Australian number that answers, finds out what the caller is after, and briefs the business.

What is true is narrower. The models this kind of software is built on keep getting better, and so do the frameworks that put them to work. Things that could not be built well until recently can be built now. We would rather be ready for that than announce it.

Sales agents and property managers do different jobs. We intend to watch that work as it is actually done, in businesses of different kinds, before we build anything more.

One thing already looks plain, and the watching will test it. The software is scattered. A listing sits in one system, an enquiry in an inbox, a maintenance job somewhere else again, the inspection notes on somebody's phone. The person doing the work carries it between them. Centralising that, across Australian real estate, is the ambition.

We think the answer is something as simple to use as a chat, with a custom framework underneath it, built to what the models can actually do and to the specific work of this industry.

We want it to fit the software a business already uses, and a working day that is mostly not spent at a desk. We have not built those connections yet. The work should take less effort, and less of it should sit behind software that is hard to learn.

Lowering the activation energy

We start from an observation, not a study: a person has a finite number of decisions in them in a day. Each one costs something, and the cost is emotional as much as it is time. It rises with the consequences. Advising a vendor to hold or to meet the offer costs more than deciding where to have lunch. So does deciding what to tell a tenant about a repair that has already waited a week.

There is a word for this in physics. A change does not happen just because the end state is better. Energy has to arrive first, to clear a threshold. That is activation energy. The picture we started from is an electron that stays put until enough energy arrives to break the pull of the nucleus. Decisions behave the same way.

The rest is a belief, and we will call it that. Put better information where a person can reach it at the moment of the decision, and the threshold drops. Lower thresholds, we think, mean more decisions made, and more decisions made is more work done. The wider bet is that if each service costs less in effort, the real estate industry in Australia moves faster.

None of that is demonstrated. We are pre-revenue, with one product and no results to point at. The belief is why we build the way we do. It is not a claim about what the product has done.

It settles one rule for the software: it does not make the decision for you. It is built to lower what the decision costs to make. The sales agent still decides what to advise the vendor. The property manager still decides what to tell the tenant about the repair. What we are trying to take away is the weight of getting to the decision, not the decision itself.

Where we will not compete

A tool that makes the job easier should not work by hollowing the job out. It is the line we draw, and it decides what we refuse to build.

Real estate runs on people who know their suburb and their clients. A sales agent who knows the street. A property manager who knows which tenant to ring first. Harbour answers at two in the morning and writes down what was said, so the person who does know is the one who calls back.

We are not trying to be the cheapest voice on the line. A call that costs less and is worse is not a saving. We think the saving comes out of local knowledge first, then out of accountability for what was said, then out of the name on the sign.

An Australian number, with an Australian company answerable for it, is a deliberate choice. It is not the cheapest way to answer a phone, and we are keeping it. It does not mean every part of a call stays onshore. Some processing happens overseas, and the privacy policy says where and by whom.

Some limits are built into the software rather than left to whoever is on shift. Harbour never rings anybody. On a call it has no way to book a time or send the caller anything, and saying it is an AI assistant is not a setting a business can switch off. The rest are rules it is told on every call: open by saying so, promise nothing except that someone will be in touch, quote only a price your notes give as advertised, and say so when it does not know rather than guess.

Harbour does not make anyone compliant. Those obligations sit with your business and they stay there. What we do is decline to build shortcuts around them, and be plain about where our software stops.

We want the people doing this work to have the best tools we can build. Not to win by making the job worth less.

Who this is for

Sales agents and property managers do not do the same job. A property manager's phone rings about a leaking tap, an overdue rent payment, an entry notice. A sales agent's phone rings about a listing, a price, an inspection time. One thing is true of both: the phone rings while you are in the middle of something else.

We are early. There is one product, and it was built for the sales enquiry. What it should do with a call about a leaking tap, rather than a listing, is one of the things we want to learn. What a useful call looks like to you. What software should never be trusted with. Where you would rather ring someone back than send a message.

If you run a rent roll, or sell houses, or both, we would like to hear how the phone actually behaves in your week. We are asking for a conversation, not a sales pitch. Write to hello@ateaseai.com.au.