2026-11-03 –, Beavis
QField, Mergin Maps, ODK: the open-source field-collection stalwarts. What happens when AI can generate the capture app from a description, or skip the form entirely and extract attributes from a photographed, narrated observation?
QField, Mergin Maps and ODK, the stallwarts of field data collection in the open source world. The form has always been the important, but often overlooked, piece in the data collection puzzle that could make or break a project. What influence will AI have on how work is done in the mapping/data collection sector.
This talk walks the spectrum. At one end, instead of hand-building a complex schema and a custom form, you describe the workflow in plain language and let your favourite LLM generate an offline-capable capture app (or even a native app for your mobile platform of choice). At the other end, you skip the form entirely: the collector photographs an asset and narrates what they see. AI then extracts the structured attributes on sync. In between sits a range of hybrids worth mapping out.
We'll take a look at where each model is today and how far each idea can go. The familiar field form might change shape, or disappear, over the next few years.
Rhys is co-founder and spatial data architect at Auchindown, a small Jamaican company that deals in all things spatial. Over more than 19 years he has analyzed, identified, and transformed spatial data into useful information and business insight, helping organizations make better-informed decisions.
He has worked with PostgreSQL/PostGIS since the days of PostgreSQL 8.0 and pre-1.0 PostGIS, and has taught Advanced PostGIS workshops at FOSS4G events around the world. His particular interest is PostGIS in the electric utility space: he is the developer behind Livewire (lvwr.io), a cloud-based power-delivery modeling and management solution built on the PRAM stack.