The GBS team from both New Zealand and the US has just returned, refreshed and full of ideas, from the Esri Partner Conference and Esri Developer Summit. Before the conferences officially kicked off, some of the team also had the chance to spend time in Redlands, California. Hard to believe it was a first visit for a few of them, especially given we’ve been delivering Esri solutions for more than 20 years.
And, in keeping with tradition, the team also made time for a visit to the legendary Las Casuelas for some not-quite-as-legendary margaritas. Photos available on request.
But back to business, here are our top 10 takeaways from the two events.
1. Data readiness and interoperability remain the biggest barriers to AI and advanced GIS
Across AI, automation, real-time GIS, and agents, the biggest challenge is still data. Quality, governance, interoperability, metadata, and clear systems of record all matter. “Garbage in, garbage out” has never been more relevant, especially when AI is involved in geospatial workflows.
2. ArcGIS is becoming a geospatial AI platform with an agent-first direction
Esri is clearly positioning agents as a core part of the ArcGIS platform, sitting between services and applications. This includes agent-to-agent interaction, reusable geo-agents, embeddings over spatial data, and agent-driven app architectures supported through MCP and ArcGIS Agent Builder.
3. Model Context Protocol, MCP, is emerging as a foundational ArcGIS capability
MCP looks set to become a standard platform capability, much like OGC or Feature Services. It will allow ArcGIS spatial data and services to be accessed by third-party agents and tools. Esri is not looking to host LLMs directly. Instead, that layer will sit in tools like Copilot Studio or ArcGIS Agent Builder, while still respecting ArcGIS Enterprise security controls.
4. AI is being treated as a core differentiator, not a side feature
AI was positioned very clearly as “the next differentiator”. Rather than being presented as an experimental add-on, it is being embedded across the platform to improve workflows through automation, scale, and better insight.
5. Responsible and trustworthy AI is a deliberate focus
Esri is taking a measured approach to AI, with a strong emphasis on trust. That includes reducing the risk of hallucinated spatial outputs, improving transparency, and introducing trust guidance and AI cards. Domain expertise and practical guardrails are still essential.
6. Assistants will be everywhere across the ArcGIS ecosystem
AI assistants are being built into ArcGIS Pro, Notebooks, Portal, field apps, and development workflows such as Arcade, Forms, and Web Maps. They are designed to be context-aware, extensible through SDKs, bundled with user types, and managed through quotas rather than token limits, all of which points to broad adoption.
7. Developer productivity and automation are moving ahead quickly
There were a number of meaningful updates for developers, including Python improvements, VS Code toolboxes, headless ArcGIS Pro automation through CoreHost, Parquet support in the JavaScript SDK, beta static map services, automated item validation, and metadata compliance tooling. The overall direction is clear, faster development and more automation.
8. Real-time GIS is becoming even more important in an AI-enabled world
Real-time GIS has long been central to ArcGIS through streams, dashboards, and web maps, but AI is increasing its value even further. As systems become better at interpreting live data, the potential for faster decisions and more responsive workflows grows significantly.
9. 3D GIS is maturing into a true system of record
Esri continues to push 3D beyond visualisation. With support for 3D object layers, Gaussian splats, voxel data, and subsurface modelling, 3D GIS is increasingly being treated as authoritative data infrastructure, not just a way to present information.
10. SDKs and APIs remain a major priority
Esri is continuing to invest heavily in SDKs and APIs, with a clear expectation that partners and developers will build fit-for-purpose tools and workflows on top of the platform. The emphasis is on extensibility, custom automation, and tailored solutions rather than one-size-fits-all products.
It’s also worth noting that, outside of the long-established GeoAI tools, most of the newer AI capabilities, particularly AI Assistants and AI Agents, are still in beta and likely around a year away from general release.