AI Chat
AI Chat is the chat assistant of Profinity AI, built into Profinity. It is available from the side menu once an administrator has configured it (refer to Profinity AI settings). It can answer questions about the live state of this Profinity instance and look up how-to and reference material from docs.prohelion.com. If the administrator has enabled it, AI Chat can also search the web.
AI Chat requires the AI licensed feature, included in the Server and Enterprise editions. Without it the feature is unavailable. Refer to Licensing for what each edition includes.
Your messages may leave this instance
An external AI provider answers Profinity AI unless your administrator has configured a local, self-hosted model. Ask your administrator which provider is configured before you discuss sensitive data with the assistant.
Open Profinity AI
Users with the Profinity AI permission see a Profinity AI entry in the side menu. Select it to open the chat window. The built-in Administrators role has the permission by default.
If the entry is missing, one of these causes applies:
- The signed-in user lacks the permission.
- The instance lacks the AI licensed feature.
- The administrator has not enabled and configured Profinity AI under ADMIN > System Configuration.
Ask a question
The chat window opens with a short greeting and three example prompts: a system-health question, a documentation search, and a how-to question. Type a question of your own, and press Enter to send it. Press Shift+Enter to add a new line without sending.
While Profinity AI is answering, the status indicator at the top of the window reads Thinking…. Once a question has been answered, it reads Connected to this environment. The response streams in as Profinity AI generates it, rather than appearing all at once. Web search, when the administrator has enabled it, is available only with the Claude, OpenAI, and OpenRouter providers (refer to Profinity AI settings).
Profinity AI answers questions about this instance by querying live data with the same permissions you have. It sees only the tags, alerts, or configuration that you can see in Profinity.
You can send another question while Profinity AI is still answering the previous one. Profinity AI queues it and answers it in turn. Use Stop to cancel a response in progress, and New chat to clear the conversation and start again.
When AI Chat reports an error
| What you see | Cause | What to do |
|---|---|---|
| The chat window shows the provider's timeout message, or "The model timed out." when the provider sends none | The provider does not answer within the Response timeout (seconds) that the administrator set | Send the question again, or ask the administrator to increase the timeout |
| "Something went wrong." followed by a link to the logs for administrators, or the text "contact an administrator for more details" for other users | Examples are an API key that the provider rejects and a provider that cannot be reached. Profinity records the cause in the logs | Administrators can follow the link to the logs. Other users contact an administrator for more details |
| The administrator corrects the provider settings under ADMIN > System Configuration. |
Limitations for 2.3
- Profinity AI keeps conversation history only for the current browser tab, and clears it when you close the tab. It does not keep a record of past conversations across sessions or devices.
- Profinity AI uses one configuration for the whole instance, so there is no per-user choice of provider or model.
- The example prompts in the empty chat window are suggestions only, and Profinity AI is not limited to those topics.
To build dashboards and other configuration with your own AI coding assistant instead, refer to AI Skills. For the Model Context Protocol (MCP) connection that Profinity AI uses for live data, refer to MCP Server.