Know what's in the data before you use it.
LumiDB now reads every scan's metadata on upload and writes an AI summary that stays with the data, for people and for the API.
Written by
Published on
Do not index
A utility receiving LiDAR data for a transmission line project will often check one number first, whether the crossings reach the density written in the specification, and many checks stop there, even though data can still be unfit for the job. LumiDB's new AI summary and metadata view gives every reader a starting point for that judgment the moment a scan is opened.
Two principles apply to any scan. First, quality is relative to purpose: data that serves a city planning model may fall short for a transmission line survey, so density, positional accuracy, classification, coverage and their consistency across the site have to be read against the intended use rather than a single benchmark. Second, quality information has to travel with the data. A report on the surveyor's drive does nothing for the person opening the project.
When that information is missing, both sides of the handoff pay for it. Too often, an engineer, a planner or a member of the public opens a project, finds that one area has different attribute values and a different level of detail from the next, with no easy way to tell why. Usually the areas were captured in separate campaigns or to different specifications, but without that context the difference causes confusion and erodes confidence in the data. Meanwhile, the survey company spends hours compiling metadata by hand from LiDAR software and several other tools, often writing reports the client never asked for, as insurance for the day that client returns needing coverage beyond the original boundary.
LumiDB now extracts that information automatically when any scan is ingested: coordinate reference system, extent, point density and spacing counted across all returns, classification distribution, flight line or scan position count, scan angle range, and the capture method recorded in the platform record. From that record it writes a short summary in plain language, so anyone with access can form a first view of whether the dataset fits their purpose before using a single point. The same record saves the survey company part of the manual compilation at closeout. It does not replace formal QA or a certified report. It is a quick first read that stays attached to the data.
The metadata record is available through the LumiDB API, so what a person reads in the summary, a script or an AI agent can query directly: which datasets cover an area, in what coordinate system, at what density, with which classes present. Metadata that used to sit in a PDF on someone's drive becomes something a pipeline can filter on before it touches a single point. The summary is one of several elements in the LumiDB handoff package, together with the project documentation and role-based instant links that give each stakeholder the right access without local setup or file transfers.
This release covers what can be read from the files and the platform record. Next, we aim to align the metadata output with the standards clients require, host checkpoint accuracy results next to each dataset and show coverage against the contracted extent. The goal is for every scan to carry what is needed to understand it, whether it is opened next week or years after capture. We are eager to hear from survey firms, asset owners, engineers, planners and anyone else who wants to read scanned data more easily.
Book a session with our team to see the summary on one of your own scans, and tell us which metric you check on every delivery.
Written by
