Skip to main content

TagoAI Chat

Use TagoAI Chat to ask questions about your TagoIO resources, investigate problems, and create or edit resources from the Admin. Before sending a message, choose the model and the permission mode appropriate for the task.

This page covers the chat in the Admin. For the assistant your portal users see, read AI Chat in TagoRUN.

Starting a chat​

Open TagoAI from the star icon in the sidebar, or from the top bar of the Analysis and Dashboards pages. You can start a new chat at any time using the New Chat button.

Each chat keeps its own history, so you can split unrelated tasks into separate conversations and come back to them later.

Choose a model​

The selected model determines the capabilities and billing path used by the chat:

Some models also let you choose a reasoning effort. Higher levels can improve results on complex tasks but can increase response time and usage.

New chats start with the profile's configured default chat model. You can select another model if it is available in the profile's chat model menu.

Page context​

When you open TagoAI from a supported Admin page, it receives context about the resource currently open. For example:

  • From a dashboard, it can identify the dashboard you are viewing.
  • From an Analysis page, it can inspect the script currently open.

This lets you ask questions such as "why is this widget empty?" or "refactor this function to handle missing payloads" without manually identifying the resource.

Expanding and retracting the panel​

The chat panel sits on the right side of the Admin. Use the expand control in the top-right corner of the panel to give the chat more room, or collapse it when you need the full screen for your work.

Chat history​

TagoAI chat panel with the chat history button highlighted

TagoAI keeps your previous chats so you can revisit what you asked before. The history panel groups chats by recency and lets you search across them.

Chat history is capped at 20 chats per profile. When you reach the limit, the oldest chat is deleted first as new ones are created. If you want to preserve a conversation long-term, copy the relevant parts out before it rolls off.

Modes​

TagoAI has three modes that control what it is allowed to do on your platform. It opens in Edit Mode. You can switch modes from the bottom of the chat panel before sending a message.

warning

In Edit Mode and Full Access, TagoAI can apply changes without asking you to confirm each action. Use Read Only first when you want to review the proposed approach before allowing changes.

Read Only​

Read Only mode only fetches information. TagoAI can list your devices, inspect dashboards, review Analysis scripts, and read data, but it will not create or change anything.

Use Read Only when you are exploring, asking questions, or want to be sure the assistant will not touch your resources.

Edit Mode​

Edit Mode allows TagoAI to edit existing resources and create new ones. It can draft an Analysis script, build a dashboard, register a device, or update an Action. Each action it takes appears in the conversation.

Edit Mode does not delete anything. It is the recommended mode for most day-to-day work.

Full Access​

Full Access adds deletion to what Edit Mode already allows. TagoAI can remove devices, dashboards, analyses, and other resources as part of completing a task.

warning

Full Access can delete resources in your profile. Deletions in TagoIO are not recoverable from the Admin, so only use Full Access when you understand what you are asking for and have a recent backup if the data matters.

AI Dashboards​

TagoAI works on AI Dashboards through their own tools, not the widget tools, which refuse AI Dashboards. Ask in plain language; for example, "create an AI Dashboard for the greenhouse sensors" or "add a 7-day temperature chart to Ana Souza's personal dashboard on Cold Rooms".

You can ask it toMode it needs
Read a dashboard's settings and rulesRead Only
List a dashboard's versions, yours or a user'sRead Only
Create an AI DashboardEdit Mode
Change the label, icon, tags, Personalized dashboards setting, or rulesEdit Mode
Build or change the layout from a requestEdit Mode
Restore a versionEdit Mode
Reset a TagoRUN user's personal dashboardEdit Mode
Create or change a TagoRUN user's access policyEdit Mode
  • Prompting. The chat sends one self-contained request to the dashboard generator, the same one the prompt bar uses, with your rules and billing. The generator sees the dashboard as it is, not the conversation. For a vague ask, the chat first looks up the devices and their variables, then writes a full request naming the devices, variables, chart types, time range, and layout. It asks you a question only when it can't tell which devices you mean. For a TagoRUN user's personal dashboard, the request names no devices: the chat's lookups use your access, so the generator picks from the user's own devices instead.
  • TagoRUN users' personal dashboards. Name a TagoRUN user and the chat prompts, lists versions, restores, or resets that user's personal dashboard instead of the main dashboard, under that user's device access. It can tell whether a user has one and whether they can edit it now. Who may edit is AI dashboard generation in Access Management, which the chat can grant by creating or updating a TagoRUN user's access policy. See Personal AI Dashboards for Run users.
  • Restore and reset. A restore is saved as a new version, so it can be undone. A reset starts a user's personal dashboard over and keeps its versions, so restoring one brings it back. The chat confirms with you before a reset.
  • What each call did. For each AI Dashboard tool call, the chat keeps what it sent and what came back. For a prompt, that is the request as sent, the generator's steps as they run (devices listed, variables read, generating, saving), and the outcome: what changed, that nothing changed, or the error.

A change made through the chat counts twice: once for the chat turn and once for the generation.

Check access rules after the chat changes them

Asked to add AI dashboard generation for a user, the chat may edit a rule that user already has, for example one on another dashboard, instead of adding a new one. After the chat changes a policy, open it in Access Management and check its rules.

Tips for better results​

  • Be specific about the device, dashboard, or analysis you want to act on. Names and IDs help.
  • Use Read Only to ask for a proposed plan. Switch to Edit Mode only after you have reviewed the plan.
  • Keep one task per chat so the history stays useful when you come back.