Many/Most of the processes in ArcGIS Pro are accessible programmatically. Look at how the various tools/processes are used from a Server / Notebook / Model point of view.
From the tone of the UC plenary (I've only seen a few minutes here & there), Esri may look into better use of agentic AI going forward.
In some aspects that are apparent from the existing AI companion use in AGOL, it seems that Esri are looking at scenarios where users 'converse' with the application to achieve the outcome they need.
From an agentic and conversational UX point of view, the developers will now need to focus on three 'UI' environments:
- the one for people to point & click as always,
- the conversational UI with more visual feedback,
- and the agentic UI where agents may need to interact with an api-like UI (piece of code runs piece of code to make dot red)
As this developmental process evolves, I suspect Esri may get into strife keeping all three streams up to date & performant. They're already not in the best spot right now and they are only focusing on one of the three (in public versions, I may be off target by a wide margin and iPro 4.A will have a spectacularly good UI and great UX).
In addition, as we evolve our own agentic solutions, I foresee the way we will solve problems will diverge significantly.
Back to the Idea.
If the Pro UI gets Open-Sourced, I hope to see that others create task-specific, or focused UI solutions or combinations that leverage the very strong toolkit and ecosystem that Esri has built.
Leaving accessing these proprietary tools and processes via a new UI to suit your project, or your personal AI agent, to others, will create significantly more opportunities for Esri to focus on the core paradigm of creating tools that solve problems, store data, find data, etc.
Going back to science fiction movies, there are many events where a person discusses a design and the AI makes changes (Iron Man for one). My take is that the key to realising this ideal is that the UI has to be built to suit conversational instruction and result-driven visual feedback.
The current Pro paradigm of tools needing focus on a particular window/pane to determine a right-click context, having dynamic toolbars or many open panels, etc. does not look to be the most suitable for the best conversational experience.
I am sure there are smart people out there that will jump on the chance to make this a reality; but it will only work if Esri can build the AI and backend enablers rather than wasting time trying to figure out why Catalog doesn't sort files & folders the same way Windows does or how to create a better GP batch processing UI that can mimic what we had in ArcMap.
This is a complex topic I find very interesting.
Kudo (or not), give feedback, and tell us how you see this evolve over time.
