From Answers to Action: Why ENVI and IDL Agents Go Beyond General AI

Zachary Norman Monday, April 20, 2026

As generative AI tools like Claude and Gemini continue to gain traction, many organizations are asking the same question:
Can general purpose AI actually support real geospatial workflows, or does it stop at surface-level answers?

That question was front and center in our recent webinar, Meet Your New Partners in Science: ENVI Agent and IDL Agent, where we explored what it takes to move from AI-generated responses to rue, executable geospatial analysis. While tools like GitHub Copilot and other chat-based platforms can accelerate coding and ideation, they often lack the deep integration, domain awareness, and operational control required for production-grade remote sensing and analytics.

ENVI Agent and IDL Agent were built specifically to close that gap. By combining large language models (LLMs) with direct access to ENVI and IDL through Model Context Protocol (MCP), these agents don’t just suggest solutions, they execute workflows, interact with your data and code, and operate within the systems you already trust.

In this blog, we compiled the most important questions from the webinar including setup, workflow compatibility, platform support, and how these agents compare to AI tools, to help you understand where purpose-built GeoAI agents deliver real value.

1. How can I access ENVI Agent or IDL Agent?

Accessing ENVI Agent and IDL Agent and requires some additional tools outside of a normal download from our website. To help, we have two short and sweet videos that walk through the process and provide quick links for each requirement:

2. How is ENVI Agent and IDL Agent different from general AI tools such as Claude, ChatGPT, or Gemini?

The primary difference of ENVI Agent and IDL Agent compared to tools like Claude Code is the native integration with ENVI and IDL.

We include dedicated sets of MCP (Model Context Protocol) tools that allow LLMs direct access to our software, which you won't find in other places. These MCP tools also enable LLMs to self-discover context based on the user problem they are trying to solve and are specific to our implementation.

3. We have complex workflows using IDL and ENVI to process data. Will these workflows be accessible by the agents or do we need to re-write them?

IDL Agent and ENVI Agent can access and work with existing code bases. One of the benefits of utilizing GitHub Copilot and VS Code is that it enables LLMs to quickly be experts at working with your code bases. No need to re-write or re-factor.

We do recommend making sure there's documentation for how to run different routines so that an LLM, just like a person, can figure out how to run your routines.

4. Can ENVI Agent and IDL Agent work with other agents or create child agents to split up tasks?

Today, ENVI Agent and IDL Agent each operate as a single, self contained agent. They don’t currently create or manage separate “child agents” the way some tools (like GitHub Copilot) can.

That said, you don’t need multiple agents to get complex work done. If you want the agent to perform multiple actions at once—for example, run different algorithms, compare results, or analyze data in several ways—you can simply ask it to do so in one request. The agent will handle those tasks for you internally and return the combined results.

5. Your webinar demo is on Windows. Is that the only supported platform?

No, IDL Agent and ENVI Agent will run wherever IDL and ENVI are supported. That means Windows, Linux, and Mac.

You can find a full list of supported operating systems for IDL and ENVI here:
https://www.nv5geospatialsoftware.com/Support/Maintenance-Detail/tag/platform-support

6. Can ENVI Agent and IDL Agent handle large, complex tasks —and stay on track while doing them?

Yes, absolutely. There's a couple of ways that this happens, based on which agent you are using. Here's some examples:

With IDL Agent and GitHub Copilot, you can start with Plan mode for the agent. In Plan mode, you can collaborate with the LLM to develop a complex, multi-step approach to solve programming problems. Once finished, the Agent can then implement your changes following a plan to keep everything on track.

With ENVI Agent, we make sure that to-do lists are created and utilized to keep the LLM on track. This is native to GitHub Copilot and is great to make sure all processing steps happen as requested and to see overall progress.

Pro tip: If you are ever worried about an LLM getting lost with a long processing workflow, you can always explicitly ask for it to create a to-do list with your first request.

7. Will other people be able to access IDL Notebooks or ENVI Modeler Workflows? Or do they need specialized software?

IDL Notebooks can be saved and shared directly with other users so they can pick up where you left off or re-use something you created. They can also be saved out as a PDF for those who don't have the ability to work with them.

For ENVI Modeler Workflows, you can share those with any other ENVI user and they can open in ENVI.

8. Can ENVI Agent work with IDL?

Yes, ENVI Agent can also write and run code like IDL Agent as it includes all of the same programming capabilities.

We made this decision deliberately because we wanted the agents to have all the tools to be able to solve problems for users. Internally at NV5, we often turn to IDL to create custom process workflows and ENVI Agent needed to be able to do the same.

9. If I already have a Claude subscription, do I need to pay for a new Copilot subscription? Is there any difference between using GitHub Copilot Pro and Pro+?

Yes, you need a separate GitHub Copilot subscription for IDL Agent and ENVI Agent to work in VS Code. GitHub Copilot includes access to Claude (and others like Codex). Check out this release with more information:
https://github.blog/changelog/2026-02-26-claude-and-codex-now-available-for-copilot-business-pro-users/

Mapping Earthquake Deformation in Taiwan With ENVI