MapInfo Pro

Welcome to the MapInfo Pro community!  Please feel free to start a discussion in the discussion tab or join in a conversation.

Here are some useful links where you can find more information:

Product Announcements

Discussions

Members

Resources

Events

 View Only

MapInfo Monday: MapInfo AI

  • 1.  MapInfo Monday: MapInfo AI

    Employee
    Posted 05-04-2026 02:05

    Happy #MapInfoMonday!

    Over the last few months, our engineers have worked on a few things related to MapInfo Pro and AI: 

    They have built an MCP server for MapInfo Pro that allows you to communicate with MapInfo Pro from your favorite AI Chat Assistant, like Claude.

    And they have also built an AI chat window that runs directly inside MapInfo Pro and connects to your Large Language Model (LLM), for example, GitHub Copilot or OpenAI.

    We have gathered this in an add-in called MapInfo AI, which is now available through the MapInfo Marketplace for MapInfo Pro v2023 or newer.

    Let me show you some of the benefits of MapInfo AI.

    Labeling using MapInfo AI

    Let’s start with a simple example. You can ask MapInfo AI to help you label a layer:

    Label the planned fiber lines with the length

    image
    MapInfo AI will check what it has to work with. It checks the available tables, maps, and layers.
    With this, it determines which layer I’m referring to.
    It now checks for existing columns that can be used and decides to use a dynamic expression for calculating the length of the polylines.
    It rounds the number nicely and adds a distance unit to the label, too.
    image
    Let me do the same for the existing fiber lines:
    Do the same for the existing
    image
    It has the context and knows what I mean, even though I’m not very specific.
    image
    For each of the commands it runs, I can click on it and see the details.
    It’s a good idea to check the statements that it uses to ensure that it’s doing it correctly.
    image

    Data Loading and Mapping using MapInfo AI

    MapInfo AI can also help you load data in various formats into MapInfo Pro and map it:

    Load and map data from this folder "C:\Maps\2025\ Fiber Planning Gainesville\Data"

    image
    MapInfo AI starts scanning the specified folder for files it can open. And then it starts importing these into MapInfo Pro.
    image
    With all the files loaded, MapInfo AI creates a map of the data with styling and some labels.
    It provides an overview of the data loaded, the object types of each table, and the layer's styling.
    image
    A nice and easy way to get started.

    Query using MapInfo AI

    Let us try with a query example. I’m looking for parcels with no buildings:

    Highlight parcels with no buildings on the map

    Notice that I told it to highlight the result, and I didn't specifically tell it to use SQL to find these parcels.

    image
    You can follow its thinking in the Chat window.
    It tries a few SQL Select statements that don’t work. Then it decides to split the process into two: Find the parcels that do have buildings and then use this to find those that don’t.
    image
    Also, notice how it gives you information about the result. Counts and details about how the result has been styled in the map.

    Create a New Map using MapInfo AI

    You can also use MapInfo AI to create a new map with either existing base tables or query results.

    Let’s create a map with a focus on the planned fiber lines, and the addresses and buildings nearby:

    Create a new map showing only the planned lines and the addresses and buildings within a distance of 250 meters from these. Use a road map as base

    image
    As before, you can follow how MapInfo AI plans and executes the work in the chat window, and in the map, layers, and queries that have been created.
    image
    I did specify to use a road map style base map, so MapInfo AI picks the Bing Road base map.
    image
    As you can see, it also provides a compressed version of the work it did, including how many buildings and addresses it found.
    It even gave the new map a friendly name, and the layers have friendly names too.
    Let’s inspect the query it used to find the buildings. A common mistake here is to use Within. This would require that the centroid of the building fall inside the buffer for the building to get selected. A better choice would be Intersects. Then, the building would get selected if any part of it falls inside the buffer.
    MapInfo AI did use the Intersects operator, as you can see in the details of the spatial joins.
    image

    Installation

    You find MapInfo AI in the MapInfo Marketplace from inside MapInfo Pro.
    image
    Click on MapInfo AI Preview under the Featured Products or find it under All Products.
    On the MapInfo AI Preview product page, click the Install Product button to download and install the add-in.
    Please note that currently, this is a preview version of MapInfo AI.
    image
    MapInfo Pro now downloads the add-in, unzips and registers it in the Tool window.

    Configuration of MapInfo AI

    MapInfo AI doesn’t include a dedicated AI account. You need an account or sign up for one. MapInfo AI can connect to this account using an API key. You can often get this API key from the configuration page of your AI.

    image
    You configure the connection via the Settings panel as shown above. Start by selecting your provider from the list, then enter your API key in the field.
    You can also select which model you want to use. This depends on the provider. Often, the newer models are more expensive than the older models, but they are also better. 
    image
    You can use the Test Connection button to check your connection to the service.
    In the next chapter, you can see how you can configure MapInfo AI to use GitHub Copilot.
    Next to the Settings button, you will find a small list that currently lets you choose between two options: Full or Limited (No MapBasic)We recommend starting with the Limited (No MapBasic) option. It controls whether you limit AI to using the specific tools we created to help it perform tasks in MapInfo Pro, or let it find solutions through the marvelous world of MapBasic. In most cases, it can solve the task without MapBasic and does so faster because there are fewer solutions to consider.
    image
    You can also try solving the same problem with both options and see which one works best. We would like to hear your findings.
    At the top of the AI Assistant window, you find the Chat History. Here you can create new chats and delete existing chats. The AI Assistant window keeps the last 10 chats.
    MapInfo AI also comes with an MCP server, allowing you to control MapInfo Pro from, for example, Claude Desktop. We will revisit this in a later article.

    Configuration of GitHub Copilot

    You need to connect MapInfo AI to a large language model. I’ll walk you through how you can connect it to your GitHub Copilot account. This requires that you have a GitHub account with Copilot access.

    Click on the AI Assistant control on the Home tab to open the AI Assistant window.

    image
    The AI Assistant window opens, offering tips on how it can help you.
    image
    To configure it, we need to click on the Settings button. This will open the Settings panel.
    image
    Select the GitHub Copilot option from the Provider listNow, click on the Login (GitHub / GitLab) button to initiate the process of getting an API key from GitHub.
    The Authorize with GitHub dialog appears. Click on the link shown in the dialog. Don’t close the dialog, as you will need the code shown there.
    image
    Select your GitHub account from the Device Activation view or use the Use a Different Account button to specify a different account.
    image
    Now you must authorize GitHub CLI for the selected account. Click on the Authorize github button.
    image
    In the Authorize your device view, you must now enter the code shown in the Authorize with GitHub dialog. Click the Continue button to get your device authorized to use your GitHub account with MapInfo AI.
    image
    You should see the next view if it all worked.
    image
    The Authorize with GitHub dialog now closes, and the API Key field in the Settings panel has been filled.
    You are now ready to start using the MapInfo AI Assistant.
    image

    We would love your Feedback

    If you get the opportunity, please try out the MapInfo AI add-in and let us know what you think.
    What did you use MapInfo AI for? How did it help you - or did it? Feel free to share your examples and your prompts.
    You can either share your feedback here in the comments section below or use Precisely Ideas.



    -------------------------------------------



    -------------------------------------------


    ------------------------------
    Peter Horsbøll Møller
    Principal Presales Consultant | Distinguished Engineer
    Precisely | Trust in Data
    ------------------------------


  • 2.  RE: MapInfo Monday: MapInfo AI

    Posted 07-10-2026 11:00

    We are having trouble with the Custom Endpoint connection. Test Connection says it's successful, but the prompts always return Error: LLM returned HTTP 404: Not Found. Have you tested any Custom providers successfully? Should I open a support ticket?



    ------------------------------
    Nick Letourneau
    Verisk
    Minneapolis, MN
    ------------------------------



  • 3.  RE: MapInfo Monday: MapInfo AI

    Employee
    Posted 07-10-2026 11:16

    Hi Nick,

    Yes, please open a technical support case via our portal.

    Please provide your steps and the full screenshot.

    Thank you in advance

    David 



    ------------------------------
    David Chevé
    Senior Support Engineer
    Precisely | Trust in Data
    ------------------------------



  • 4.  RE: MapInfo Monday: MapInfo AI

    Posted 30 days ago

    Hi,

    My employers AI policy is to only use Microsoft Copilot while signed in to our Microsoft account.

    This gives protection to any information being shared publicly which is important to them.

    There doesn't appear to be an option (last time I checked) to use Microsoft Copilot with MapInfo AI.

    Are you able to confirm this?

    Can you provide any advice I can give to my employer that they may allow me to use other providers while not allowing public access to any information I provide?

    Regards

    Mark



    ------------------------------
    Mark McGovern
    Meander Valley Council
    ------------------------------



  • 5.  RE: MapInfo Monday: MapInfo AI

    Employee
    Posted 30 days ago

    Hi Mark,

    Copilot can't be used alone. You need to have GitHub Copilot, as that gives you API access. Your company needs to subscribe to a GitHub plan with Copilot.

    Regards

    David 



    ------------------------------
    David Chevé
    Senior Support Engineer
    Precisely | Trust in Data
    ------------------------------



  • 6.  RE: MapInfo Monday: MapInfo AI

    Posted 25 days ago

    Thanks David. I appreciate the reply.

    Regards

    Mark



    ------------------------------
    Mark McGovern
    Meander Valley Council
    ------------------------------



  • 7.  RE: MapInfo Monday: MapInfo AI

    Employee
    Posted 26 days ago

    Hey Mark

    I got this content from our engineers. It has been generated through AI. Maybe you find this useful.

    1. Can MapInfo AI use Microsoft Copilot?

    Confirmed: no, and not for a MapInfo limitation reason.
    The providers the add-in supports are, per `Chat/McpChatViewModel.cs`:
    • Ollama (Local)
    • GitHub Copilot
    • OpenAI
    • Anthropic (Claude)
    • GitLab (Token)
    • AWS Bedrock
    • Custom (any OpenAI-compatible endpoint)
    Microsoft 365 Copilot / Copilot Chat does not expose a chat-completions API that a third-party desktop application can connect to. It is a first-party experience (web, Office, Teams, Windows). There is no endpoint or API key that an application like MapInfo AI could point at, so this cannot be added by MapInfo, it's a Microsoft platform constraint.

    2. The Microsoft-sanctioned alternative: Azure OpenAI

    Your employer's actual requirement is almost certainly "data must stay inside our Microsoft tenant and must not be used for public model training." Microsoft's product for exactly that is **Azure OpenAI Service** (now Azure AI Foundry), and it does expose an OpenAI-compatible API, which MapInfo AI can use today via the Custom provider.
    Points worth making to your employer about Azure OpenAI:
    • It runs inside your own Azure subscription and tenant, under your existing Microsoft enterprise agreement and data-protection terms, the same commercial data protection that covers M365 Copilot.
    • Prompts and responses are not used to train or improve any models, and are not shared with OpenAI.
    • You choose the Azure region, so data residency (e.g. UK South, Australia East) can be guaranteed.
    • Access is controlled by Microsoft Entra ID, so it's governed by the same sign-in, conditional access, and audit policies as the rest of your Microsoft estate.
    • Traffic can be locked down with Private Endpoints / VNet integration so nothing traverses the public internet.
    • Usage is logged and auditable in Azure Monitor.
    To configure it in MapInfo AI: select Custom as the provider and set the endpoint to your deployment, e.g. https://<your-resource>.openai.azure.com/openai/deployments/<deployment-name>/chat/completions?api-version=2024-10-21
    One caveat to flag to whoever sets this up: the add-in sends credentials as an Authorization: Bearer <token> header. That works with Microsoft Entra ID tokens; if your Azure OpenAI resource is configured for api-key header authentication instead, that variant isn't currently supported. If that's a blocker for you, let me know and I can add an Azure OpenAI provider option with api-key support.

    3. The zero-egress option: Ollama (Local)

    If the policy is absolute "no data leaves the machine at all", the Ollama (Local) provider runs an open-weights model (Llama, Mistral, Qwen, etc.) entirely on your own PC. Nothing is transmitted anywhere. Quality is lower than Claude/GPT-4-class models, but for many MapInfo tasks, it's serviceable, and it's the easiest option to get security sign-off for.

    4. General points that usually satisfy an AI policy review

    • AWS Bedrock offers the same guarantees as Azure OpenAI (no training on your data, regional isolation, VPC endpoints, IAM-governed) if your organisation is AWS-aligned rather than Azure-aligned. MapInfo AI supports it natively.
    • Anthropic and OpenAI commercial/enterprise API tiers also contractually exclude API data from model training by default. This is a common misconception, as it differs from their consumer chat products.
    • You control what is sent. MapInfo AI only transmits your typed prompt plus the metadata the tools return (table names, column names, row counts, and so on). It is not uploading your data files wholesale. It may also send some sample records. though.If your sensitivity concern is about the contents of specific tables, that can be scoped by how you use it.
    • Credentials stay local. API keys/credentials are stored in your own roaming profile, not shared with MapInfo or Precisely.
    Short version for your employer: Microsoft Copilot itself can't be integrated by any third-party application, but Azure OpenAI is Microsoft's own enterprise-grade equivalent, carries the same tenant-bound data protections, and works with MapInfo AI today via the Custom provider. If they want zero egress, use the local Ollama option instead.


    ------------------------------
    Peter Horsbøll Møller
    Principal Presales Consultant | Distinguished Engineer
    Precisely | Trust in Data
    ------------------------------



  • 8.  RE: MapInfo Monday: MapInfo AI

    Posted 25 days ago

    Thanks Peter.

    I'll pass this on. Very much appreciated.

    Regards

    Mark



    ------------------------------
    Mark McGovern
    Meander Valley Council
    ------------------------------



  • 9.  RE: MapInfo Monday: MapInfo AI

    Posted 18 days ago

    How does one connect using IAM roles? I have a support ticket open, and they sent a screen shot with a Connection screenshot that includes other fields that I don't have. I don't see any way to get them. This is what I have:

    This is what they sent:

    image.png

    How do I get to this Connection screen?



    ------------------------------
    Nick Letourneau
    Verisk
    Minneapolis, MN
    ------------------------------



  • 10.  RE: MapInfo Monday: MapInfo AI

    Employee
    Posted 18 days ago

    Hey Nick

    Can you check if there is a newer version of MapInfo AI in the MapInfo Marketplace?

    If there is, click the Update Product button to download and install the latest version.

    image

    Cheers

    Peter



    ------------------------------
    Peter Horsbøll Møller
    Principal Presales Consultant | Distinguished Engineer
    Precisely | Trust in Data
    ------------------------------



  • 11.  RE: MapInfo Monday: MapInfo AI

    Posted 18 days ago

    That is the version I have. 



    ------------------------------
    Nick Letourneau
    Verisk
    Minneapolis, MN
    ------------------------------



  • 12.  RE: MapInfo Monday: MapInfo AI

    Employee
    Posted 16 days ago

    Hey Nick

    Not sure what version support have. I have the same properties for AWS Bedrock as you do.

    Could they be running a not-yet-released version?

    Peter



    ------------------------------
    Peter Horsbøll Møller
    Principal Presales Consultant | Distinguished Engineer
    Precisely | Trust in Data
    ------------------------------



  • 13.  RE: MapInfo Monday: MapInfo AI

    Posted 16 days ago

    Yes, that turned out to be the case. Support told us it is in the works, and the update is expected to be uploaded shortly, but no release date is set. We were also told we could connect with IAM roles, however we have not been told how to do that since the current version is asking for an API key.



    ------------------------------
    Nick Letourneau
    Verisk
    Minneapolis, MN
    ------------------------------