Chapter 07
The category boundary in practice
How RDI differs from cameras, AI dashboards, and BIM coordination — and how to keep the distinctions clear in procurement and operations.
01
What this lesson is about
Procurement conversations about RDI are often muddied by adjacent categories. Camera vendors describe themselves as RDI platforms. AI dashboard vendors describe themselves as RDI platforms. BIM coordination platforms describe themselves as RDI platforms. None of those claims is malicious; each category is genuinely useful, and each touches part of the RDI stack. By the end of this lesson you should be able to walk into a procurement meeting and place every vendor on the right layer, without becoming the person who corrects categories all afternoon. The category boundary matters because the questions a buyer asks are different at each boundary, and getting the questions right protects the project from buying a layer when it needs the stack.
02
Versus cameras
A camera is a sensor. RDI is a discipline. A camera produces footage, which is the input layer of RDI rather than the output. Camera platforms have evolved to include playback, search, and basic timeline tools, all of which are real value at the capture and retrieval boundary. They are not the closed loop. The honest test for procurement is operational: can the buyer name the workflows the platform completes, and can the platform produce closeout records that an external reviewer would accept. A camera vendor will describe the viewer; an RDI platform will describe the loop. The difference is rarely visible from the marketing and almost always visible from the workflow walkthrough. Ask the camera vendor to walk you through how a delay claim closes from trigger to closeout record. The walkthrough either exists or it does not, and the gap is the category boundary.
03
Versus AI dashboards
AI dashboards belong to the interpretation layer of RDI. They surface patterns and exceptions, which is genuinely useful work. They are not a substitute for the layers below or above. A dashboard without ground truth cannot defend its findings; the moment the finding is challenged, the dashboard owner is back in the footage hunting for context. A dashboard without action routing produces sophisticated views that the project enjoys for two months and ignores by month four because the loop never closes. The right relationship is interpretation feeding action and command, not interpretation as a complete answer. The procurement question for the dashboard vendor is which workflows the dashboard completes, not which detection models it runs. The vendor that answers the first question is operating in the RDI category; the vendor that answers only the second is selling a layer.
04
Versus BIM
BIM coordinates design intent. RDI verifies construction reality. The artefacts are different: the design model versus the as-built record. The two are complementary, not competing. A mature project runs both. The model coordinates intent before the work happens. The reality data verifies the as-built record after. The design verification workflow joins them by checking installed work against the issued model, surfacing deviations early enough to be cheap rather than late enough to be claims. Either tool alone is incomplete. The pair is what mature owners ask for. The procurement question for the BIM vendor is how the as-built record flows back into the model and back into the handover pack; for the RDI vendor, the question is how the workflow surfaces deviation against the model. The category boundary is sharpest in procurement and clearest in operations.
05
How to run a procurement meeting that keeps the boundaries straight
A practical procurement meeting starts with a one-page diagram of the stack and the four boundaries. Each vendor places themselves on the diagram before the demo begins. The room then asks two questions of every vendor regardless of category: which workflows do you complete, and what does the closeout record look like. Vendors who do one part of the stack well will answer for that part and acknowledge the rest. Vendors who claim the whole stack will struggle to answer for the parts they do not own. The aim is not to embarrass vendors. It is to make the project’s buying decision clearer: are we buying capture, interpretation, BIM coordination, or a closed loop. Most projects need a combination. The boundary matters because the combination needs to fit together, and that requires the project to know what it is buying at each layer.
Practice
01. Take your three current technology vendors. For each, place them on the RDI stack and write a single sentence on the workflow they complete.
Look for: A strong response is honest about which layer each vendor occupies and names the closed loop they actually deliver rather than the one they market.
02. Draft the two procurement questions you will ask every vendor in your next round, designed to surface the workflow rather than the feature.
Look for: A strong response phrases the questions in operational language: which workflows close, and what does the closeout record contain.
Checkpoint
For your most recent vendor conversation, which side of each boundary did the vendor sit on?
Recommended reading