Search that comes with the model
Every published scope has search over its data, generated from the model and kept in step with every change. Here is what a user gets, then what you no longer have to build.
What a user gets
Text search
Ranked by relevance and tolerant of typos: "blomsterhset" finds Blomsterhuset. The matched words are highlighted.
Records like this one
Similar listings, related articles.
One hit per group
The best listing per area, one product per brand.
Places
An entity with an address is located automatically and kept current. Nearest first with the distance, within a radius or a map box, clusters on a map.
Nearest first: Leksand 19 km, Rättvik 32 km, Mora 51 km.
Saved searches
"Tell me when a new match appears." Delivered as an event to an application, or into an AI agent's inbox through the MCP server.
Aggregations
Counts, sums, averages, statistics, percentiles, histograms, per day, week or month, top records per group.
Filters and sorting
On any property, reaching into related entities.
Always in step
Every index is kept in step with every change, automatically. Nothing to rebuild.
Everywhere, governed
All of it through the API, the SDKs and the MCP server, under the scope's permissions.
What you no longer build
No Elasticsearch project, no indexer, no delta jobs, no geocoding job, no alert job. Dataceen generates them.
Seen in testing, on demo models
A country-house listing model answered "nearest first", fuzzy text, and a saved search for "under three million" that collected exactly the right new listings.
A pet-clinic model and a bike-workshop model got six working applications, one per language, with nearest-first lists, maps and fuzzy search.
CountryHouses and PetClinics are invented demo companies, not customers.