Esri News Roundup, from dymaptic

AI-generated, human-reviewed. Scout, our AI content assistant, drafts each edition from material the dymaptic team gathers: case studies, software news, and industry insights from Esri's blog and community, social media, Esri emails, and notes from our Esri Technical Advisor. We review before sending, but verify anything critical against the linked sources.

Esri is putting more prepared context and more conversational access into the places you already work, from a new demographic data layer in Living Atlas to chat-driven map apps and AI agents that can call location services directly.

Quick Summary

• Ready to use: Survey123 Mobile and Survey123 Studio are the next-generation versions of the Survey123 field app and Connect authoring tool, and Esri supports them side by side with the 3.x apps, so field teams can migrate one project at a time.
• Ready to use: USA Geodemographic Embeddings is a free Living Atlas layer that boils thousands of demographic, housing, and environmental variables down to a short numeric summary for each small area, and it suits analysts doing site selection or "find me places like this one" work.
• Worth a pilot: The Waze for Cities Live Alerts Data Pipeline brings near real-time Waze incident reports into ArcGIS with filtering and history, and it fits transportation and public safety agencies already in the Waze for Cities program.
• Worth a pilot: The Google Photorealistic 3D Basemap in ArcGIS Online covers more than 2,500 cities and needs an administrator to turn it on, so it suits planning and public-works teams with a specific visualization need.
• Worth watching: Data Explorer is a beta Instant Apps template that lets people ask questions of a web map in a chat box, and it requires complete layer metadata plus administrator-enabled AI capabilities.
• Worth watching: Beta Model Context Protocol support in ArcGIS Location Platform lets AI assistants call geocoding, routing, and elevation services, and it suits developers experimenting with AI tooling today.

Ready to Use

Survey123 Mobile and Survey123 Studio. Esri has released Survey123 Mobile and Survey123 Studio as the next-generation successors to the current Survey123 field app and Connect authoring tool, arriving alongside the platform's tenth anniversary. The new apps are built on a modern architecture that Esri says is designed for faster innovation.

The changes that will show up first in daily work are practical ones. Tabular repeats let you manage related records in a grid rather than one nested panel at a time, which matters for anyone collecting multiple readings, assets, or specimens under a single parent record. Mapping capabilities have been enhanced, and QuickCapture integration is now part of the picture, so rapid point collection and detailed form entry are less separate than they used to be. There are also AI-powered assistants for building forms and translating them, which cuts down on the manual work of maintaining a second-language version of a survey by hand.

The transition detail is the reassuring part. Esri is moving to the next-generation apps gradually across quarterly releases as feature parity fills in, and both the 3.x apps and the next-generation apps remain supported side by side. You migrate existing projects at your own pace rather than on a schedule someone else set. For a utility or public health group running dozens of live surveys, that means you can move one low-risk form first, confirm your data flows and your Connect-authored logic behave the way you expect, and leave the rest alone until you are ready. Start with a form that has a simple schema and a small field crew, and use it to learn where Studio's authoring differs from Connect before you touch anything mission-critical.

USA Geodemographic Embeddings. Esri has published USA Geodemographic Embeddings in ArcGIS Living Atlas, covering the contiguous United States and Alaska. Each small area gets a list of 256 numbers that summarizes thousands of demographic, socioeconomic, housing, and environmental variables at once. The areas are H3 hexagon cells at resolution 7, which works out to roughly 5 square kilometers each, so a mid-sized city is covered by a few dozen cells.

What this lets you stop doing is the variable-picking argument. Site selection and similarity work usually starts with someone assembling a list of twenty candidate variables, defending each one, and normalizing them by hand. Esri's companion post, an introduction to embeddings for GIS analysts, walks through four applications: finding comparable locations, clustering to surface regional patterns, feeding the numbers into a prediction model as extra input, and reducing them down for exploratory visualization. Esri built the layer using a Geodemographic Foundation Model, which means the model learned on its own how to compress many variables into a short list of numbers that captures how places resemble each other.

Esri is straightforward about the tradeoff: the compressed summary is harder to interpret than a named variable, because its internal logic is not easy to explain in plain terms. If you need to tell a city council exactly why one census tract scored higher than another, a traditional variable-by-variable model is still the one you can defend line by line. Where the embeddings shine is the exploratory pass: retail and health-system analysts screening hundreds of candidate sites, or planners looking for neighborhoods that resemble one that already works, can narrow the field fast and then build a transparent model on the short list.

Worth a Pilot

Waze for Cities Live Alerts Data Pipeline. Esri has released the Waze for Cities Live Alerts Data Pipeline, which pulls accidents, hazards, jams, and road closures from the Waze Partner Hub API into ArcGIS in near real time. You control update frequency, filter which alert types you keep, and retain history, which the earlier read-only Marketplace layer did not allow. Esri notes the Marketplace layer will be retired in December 2026. History retention is the quiet win here, because it turns a live map into something you can analyze: a traffic engineer can look at where jam alerts cluster by hour of day rather than only at what is happening right now. This suits transportation departments and emergency operations centers already enrolled in Waze for Cities, and the first setup task is getting your Partner Hub API access sorted.

Google Photorealistic 3D Basemap in ArcGIS Online. The Google Photorealistic 3D Basemap has been in beta in ArcGIS Online since February 2026, and it works in Scene Viewer, Instant Apps, and Experience Builder. Coverage spans more than 2,500 cities across more than 49 countries, and you overlay your own layers on top without writing code. Two prerequisites shape who should try it. An administrator has to enable it in organization settings, and usage is measured in sessions, so plan for the consumption alongside your other credit-based services. Urban planning, disaster response, and construction visualization are the use cases Esri highlights, and the teams that get the most from a pilot are the ones with a specific public meeting or design review coming up where a recognizable 3D view saves a long explanation.

Worth Watching

Data Explorer. Data Explorer is a new Instant Apps template in beta that lets people ask questions of a map in plain language. Three AI agents handle the work behind the chat box: one explores the data, one handles navigation, and one provides help. Instead of teaching a public audience how to use a filter panel, you give them a text box.

The setup requirements are real. Your feature layers need complete metadata, because the agents rely on field descriptions to understand what they are querying. You generate embeddings for the web map, and an administrator has to enable AI capabilities for the organization. For Data Explorer to suit a broader set of teams, most organizations would need their layer metadata in better shape than it currently is. That is a useful project on its own merits, and it is the thing to start on now if a conversational app is somewhere on your roadmap.

Model Context Protocol support in ArcGIS Location Platform. ArcGIS Location Platform now has beta support for Model Context Protocol, a standard way for AI assistants to discover and call outside services. In practice, an assistant can geocode an address, route between points, look up elevation, and produce a static map, and it can chain several of those calls together to answer one multi-step request. Esri's developers lounge post notes that access runs through the Early Adopter Community and that GitHub Copilot, Claude, and AWS Bedrock are among the supported clients, with GeoEnrichment planned. MCP support carries no charge beyond standard service consumption, so the cost of experimenting is whatever the underlying calls would have cost anyway. The same release also confirms the ArcGIS Static Maps Service reached general availability in July 2026, which is the piece you can build on today. Developers already using an AI coding assistant are the natural first group to try this.

New in the Developer SDKs

ArcGIS Maps SDK for JavaScript 5.1. The 5.1 release arrives with no breaking changes. Editing gains M and Z value support, tangent curves, and multiple-choice inputs, and there are six new map components plus AI components including a Knowledge Agent and an Arcade assistant. Parquet layers get spatial optimization for better performance, and the Arcade Editor now supports debugging. Dymaptic builds GeoBlazor on this SDK, so the editing improvements and the new map components are changes we track directly for GeoBlazor.

What This Issue Does Not Cover

This roundup is a curated selection, not a complete list of every Esri and industry release. If something important isn't here, it may simply not have made the cut this edition.

About Dymaptic

Dymaptic is an Esri Gold Partner building geospatial AI tools that extend the ArcGIS platform with grounded, trustworthy AI capabilities. From custom software to AI agents that work alongside GIS analysts, dymaptic helps public sector teams and mission-driven organizations turn spatial data into clear, defensible answers.

Generated using Anthropic's Claude Sonnet 4.6 on September 3, 2026; Reviewed by E.Brooks