How to Choose an Augmented Reality Company

Quick answer: The decision that drives every quote is who builds the 3D model. Vendors who assume you supply it quote low and correctly, vendors who build it quote higher and include the work, and comparing the two without noticing is how augmented reality projects end up half funded. Ask that first, then ask how placement is anchored.

There is a reliable pattern to how augmented reality procurement goes wrong, and it does not involve anybody behaving badly.

A project team writes a brief. Three vendors respond. The quotes differ by a factor of five or ten. The team assumes the expensive one is padded or the cheap one is inexperienced, picks somewhere in the middle, and discovers halfway through that the model nobody budgeted for does not exist.

Every vendor answered the brief honestly. They just answered different briefs, because the brief left one thing unstated.

The question that has to come first

Who builds the 3D model.

Every augmented reality experience displays a model. Platforms ingest one. Apps import one. Agencies expect one to be delivered. None of them create one from a set of drawings, and that creation is the largest single cost in most real estate augmented reality projects.

So before comparing vendors, establish what you actually have. A coordination model built for construction is not a display model and will not run on a phone. A folder of renderings is images rather than geometry. A set of drawings is the starting point for modelling rather than an input to an app.

Once that is established, quotes become comparable, and usually the expensive one turns out to be the only complete one.

The second question: how is it anchored

Ask how the model is placed on the site, and listen for whether the vendor volunteers the word drift.

Phone augmented reality anchors by tracking visual features in the camera feed and estimating its own motion. In a textured interior that is reliable. On an open, flat, low contrast parcel it degrades, and a model that looked anchored at the start of a walk can sit visibly off its footprint by the end of one.

A vendor who acknowledges this and explains how they mitigate it, by planning viewing positions, by using site markers, or by moving to a surveyed system, is a vendor who has done this outdoors. A vendor who says accuracy is not a concern has done it in an office.

The third question: who is going to be standing there

Augmented reality only works for people physically at the location. That is not a limitation to work around, it is the definition of the format, and it makes the audience question decisive rather than incidental.

If the audience is out of state buyers, remote investors or a lender in another city, augmented reality delivers nothing to them and the budget belongs in formats that travel.

If the audience is neighbours, walk in prospects at a sales trailer, a community meeting or somebody being shown around the site, augmented reality does something no other format can.

Any vendor who does not ask who the audience is, and where they will be, is not scoping the job.

How this list was put together

Entries were identified through public research and are reachable products, platforms or services. They span platforms, agencies, application developers, surveyed systems and production services because those are the genuinely different kinds of company a buyer will encounter, and telling them apart is most of the decision.

Criterion	What we looked for
Company type	Platform, agency, application developer, hardware system or production service.
Model assumption	Whether the quote expects you to supply usable geometry.
Placement approach	Visual tracking on a phone, or surveyed coordinates.
Audience fit	Public walk up, sales team, field and trade, or investor.
Stated limits	Where each option stops being the right answer.

Editorial note: Rendimension publishes this guide and appears on it. We place a platform first because a planning data layer and a production service are not competing purchases, and we list ourselves in the narrow niche we serve rather than at the top. Every other entry is an independent company we do not control, identified through public research.

1. inCitu

A useful first reference point when choosing, because it shows what a mature deployment looks like and sets a standard to compare vendors against.

The platform converts planning data into a full scale augmented reality layer viewable by scanning a QR code at the location, with published work across more than four thousand construction sites and municipal collaborations including the Washington DC Department of Buildings and a city wide layer in Charleston, South Carolina.

For a buyer choosing a vendor, the instructive detail is the delivery mechanism. A code at the site opening in a browser is a deliberate choice about adoption, and any vendor proposing an app download for a walk up public audience should be asked why.

It is a platform working from existing planning data rather than a production service, which is exactly the distinction the rest of this page is about.

2. Rendimension

Second, in the niche we serve: producing the augmented reality preview itself from drawings, for projects that do not have a display ready model.

The reason to include a production service in a page about choosing a company is that the choice is usually miscast. Buyers believe they are choosing between augmented reality vendors. In practice they are choosing who does the modelling, and every price difference follows from that.

A vendor who assumes you supply the model quotes low and is not wrong to. A vendor who builds it quotes higher and includes the work. Neither is dishonest, and comparing them side by side without noticing the difference is how projects end up half funded.

We start from construction documents because that is the state most projects are in, and we say when augmented reality is the wrong purchase, which happens often enough to be worth stating in a sales conversation rather than after one.

Declared terms rather than claims: first visuals in 48 to 72 hours, and reasonable revisions are included at no extra charge. We do not raise capital, we do not obtain approvals and we do not guarantee them, and we do not sell or lease property.

3. Rock Paper Reality

An augmented reality design agency working with startups and large brands, representative of the largest and most capable segment of the augmented reality industry.

Agencies of this type are genuinely expert at shipping augmented reality that works on consumer hardware at scale, which is harder than a demonstration suggests.

The question to ask is about domain rather than capability. Brand and retail augmented reality assumes the client supplies the asset, because in that world the asset is a product model that already exists. In real estate it usually does not, and a scope written on the brand assumption will exclude the largest line item.

4. R2U

A proptech company producing immersive sales experiences for residential developers, including augmented reality generated from architectural plans.

Worth evaluating specifically as a productised option rather than a bespoke one, because the choice between those two structures matters more than the choice between individual vendors.

A productised platform amortises across many similar units and gives a sales team a consistent tool with predictable cost. It fits volume residential well and fits a single distinctive building poorly.

5. Trimble SiteVision

Evaluate this one against a single question: will anybody make a decision based on where the model appears to sit.

SiteVision places a design model on the physical site by surveyed GNSS coordinate at 1:1, and the 2026.10 release brings lidar scan data into the same Trimble Connect workflow.

If the answer to the question is no, and the purpose is purely to communicate scale and character, then this is more instrument than the job requires and a phone based experience is sufficient.

If the answer is yes, then no amount of visual polish substitutes, and a vendor who cannot discuss positioning accuracy is not the right vendor for that job.

6. CitrusBits

A development company offering augmented reality and XR application work alongside mobile development, and the vendor profile most search results will surface.

The evaluation question here is what the deliverable actually is. An application, a branded experience and a piece of software are all reasonable things to buy. A convincing building is a different purchase that happens to be displayed through one of them.

Ask for the split explicitly. Application development, content production and ongoing maintenance are three separate cost centres and a single line item hides which one is being quoted.

7. ARki

The low commitment way to find out whether augmented reality helps on a specific project before hiring anybody.

ARki imports FBX from the tools architectural teams already use and is built for showing work in person, with capture for sharing afterwards.

Running a rough test in house first is the most underrated step in this whole process. It costs very little, it answers whether the site and the audience suit the format, and it converts a vague brief into a specific one before any vendor quotes against it.

Warning signs in a proposal

No question about the model. If a vendor quotes without asking what geometry exists and in what state, they have priced a job they have not scoped.

A single line item covering everything. Application development, content production, model preparation and maintenance are separate costs, and a merged figure hides which of them is missing.

No mention of device performance. Augmented reality that runs on the newest phone and stutters on a three year old one has failed for a meaningful share of the audience.

Ownership left vague. If the contract does not say who owns the model afterwards, assume you do not, and price the next phase accordingly.

No discussion of what happens if the tooling changes. Adobe Aero reached end of support on 6 November 2025, and it was a mainstream tool from a major vendor. Anything built on a single proprietary app carries that risk.

Questions worth asking directly

What state does the model need to be in before you start, and what do you do if we only have drawings.

What polygon budget do you target, and on which reference device.

How do you handle placement on an open site, and what drift should we expect over a fifty metre walk.

Does the viewer need an app, or does a link open it, and why did you choose that.

Who owns the model afterwards, and can we reuse it for renderings, a tour or a later phase.

What does maintenance cost in year two, and what happens if the platform changes.

When the right answer is to not buy augmented reality

A good vendor will say this and it is worth knowing the cases in advance.

The audience is remote. Nothing about augmented reality reaches somebody who is not there.

There is no site access. Without somewhere to stand the format has no subject.

The budget covers one visual asset. That asset should be a rendering, because it reaches everybody.

The decision at stake is dimensional and the budget is phone based. That is a measurement question being answered with a persuasion tool, and it is the one failure mode with real consequences.

Frequently asked questions

Why do augmented reality quotes vary so much?

Almost always because of an unstated assumption about who builds the 3D model. Vendors who expect you to supply it quote for display work only, while vendors who build it from drawings include the largest cost. Both can be honest quotes for different scopes.

What should we ask first when contacting a vendor?

What state the model needs to be in before they start, and what happens if you only have drawings. That single question separates vendors who have scoped the job from vendors who have priced a fragment of it.

How do we know if a vendor has done this outdoors?

Ask how placement is anchored and whether they raise drift without prompting. Anybody who has run augmented reality on an open, flat site knows visual tracking degrades there and will explain how they mitigate it. Anybody who says accuracy is not a concern has worked indoors.

Should the experience require an app download?

For public or walk up audiences, no. A link or QR code opening in the browser removes the barrier that most often kills adoption. A dedicated app makes sense for a sales team using it daily, where heavier models and offline use justify the install.

Who should own the model when the project ends?

You should, and it should be written down. The model is the durable asset that produces renderings, animation, tours and future phases. A narrowly licensed model means paying to rebuild the same geometry at the next stage.