Elastic Features Anyone.com: The Engineering Behind Your Next Home Move

Anyone.com now holds 210 million production records on a single Elasticsearch index. In one internal benchmark, a country-wide count fell from about 6.5 seconds to 5 milliseconds at the median.

Elastic has published a customer story about the engineering behind Anyone.com. It gives a look inside the technology supporting our global property marketplace.

For homebuyers and sellers, that work has a practical purpose. You need information that helps you compare your options, understand a property and decide what to do next. The platform has to make that information useful while handling the scale and complexity of real estate across countries.

The search box is the part you see. Behind it sit data pipelines, quality checks, ranking systems and an infrastructure built to grow. That foundation is also central to our upcoming AI transaction experts.

What happens when you move the map

Imagine you’re looking for a home within a particular budget. You widen the area, change the number of bedrooms and then narrow your search again.

Each adjustment asks the platform to do several things: identify the relevant area, apply your requirements, find matching records and organise the results. The total shown beside your search is another calculation.

A search engine can combine location queries, filters and calculations across results. The engineering challenge is making those operations work together as the amount of information grows.

For a buyer, this matters because finding a home is usually a process of comparison. You learn what your budget buys, decide which requirements are essential and explore alternatives. Search needs to support that process as your priorities become clearer.

For a seller, the same search machinery helps connect a property with people whose requirements it fits. How information is organised, retrieved and presented becomes part of how a home can be discovered.

A worldwide marketplace begins with the data

Address formats, measurement conventions and the property details available differ between countries and sources. A platform has to organise that information into records it can work with consistently.

Our data arrives from dozens of sources. Apache Airflow coordinates ingestion, with qualifying records entering Elasticsearch after quality checks. The wider infrastructure spans Google Cloud and AWS.

That is a lot of work before a home appears in a set of results.

A buyer might see a price, location and property description. Underneath, a system needs to know how to apply a price limit, place the home in the relevant search area and retrieve the details needed to compare it with alternatives.

For sellers, providing clear property information matters for the same reason. The more clearly a home’s characteristics can be understood, the better the starting point for connecting it with a buyer’s requirements.

Useful property search depends on both the information and the engineering that makes it usable.

Rebuilding the search while preserving its behaviour

Our previous Postgres setup had reached limits within our application architecture. Our team rebuilt the query layer in about a week and a half, preserving the customer experience. That included moving weighted real estate agent recommendation scores into the search engine and reproducing them accurately.

This is a less visible part of building a serious property platform: checking that a change underneath the product preserves the behaviour people rely on.

A customer should be able to understand what a filter does. Recommendations should follow their intended criteria. Growing the platform creates an obligation to maintain that consistency.

That attention matters because people make consequential decisions using the information in front of them. Our view is that technical progress has to serve the person using the platform.

Why fast retrieval matters during a home move

The 5-millisecond result measures a specific counting operation. It shows progress in a part of the search system that has to handle large sets of records.

For buyers and sellers, the wider value is what efficient retrieval can support.

Consider preparing an offer. You may want to revisit a property’s details, compare alternatives and understand the information behind a price discussion. A seller may need to review the context around their asking price or respond to a buyer’s questions.

An AI expert supporting those decisions needs to retrieve relevant information repeatedly as the conversation develops. A different budget, a new property or a change in priorities can create another information request.

Fast access to the relevant data is one of the foundations for useful guidance throughout a transaction.

The foundation for your AI transaction experts

Today, our production search uses keyword retrieval. Semantic retrieval and retrieval augmented generation, or RAG, are on the roadmap for our agentic system.

Semantic search helps a system find information by meaning. That is useful when people describe their needs in everyday language and the relevant property information uses different words.

RAG connects an AI model with information retrieved from a data source. For a property conversation, the purpose is to give the system relevant material to work with when it prepares its response.

These are practical engineering requirements for the experience we want to offer.

We’re preparing to bring digital buyer and seller brokers to buyers and sellers in 26 countries.

For buyers, the aim is an expert that helps them find suitable homes, understand pricing, prepare for negotiations and navigate the steps towards closing.

For sellers, the aim is support with presenting their property, understanding pricing, handling offers and progressing the sale.

The ambition is to make expert guidance more accessible across the journey, while giving people more control over their own transaction.

A platform you can already use

Anyone.com already gives homeowners the tools to sell directly or work with a real estate agent. You can publish your property, communicate with interested buyers, organise viewings, receive offers and manage the sale in one place.

The upcoming AI experts will build on that experience.

We’re sharing the engineering because the infrastructure behind a property platform deserves attention. When you’re choosing where to manage a home move, the work behind the interface matters.

Our goal is to make that work useful in the moments you need it: finding the right home, understanding your options and taking the next step with confidence.

Read Elastic’s full story about Anyone.com, or explore Anyone.com to start your next move.