Getting Started with Connext AI

Sign up

A Connext AI account is required to use the Connext Chatbot, the Connext Copilot extension, the AI features in System Designer, and the Connext MCP Server. If you already have an RTI Academy or Support Portal account, you can login using your existing RTI credentials; otherwise, create a new account.

To create a free account, go to the signup page.

Connext AI Suite

Connext Chatbot

To log in, go to the Connext Chatbot page.

Once logged in, enter your query, instruction, or request into the chat interface to start a conversation. You can ask for specific information or get coding assistance on all Connext products and features. See Prompt Examples below for tips on how to get the best responses from Connext AI.

Connext Copilot VS Code extension

To download the Connext Copilot extension, go to the Visual Studio Code Marketplace.

Refer to the extension's documentation for installation instructions and usage details.

Connext MCP Server

Use the following connection details:

Client-specific setup references:

In VS Code, you can use the Command Palette to add a new MCP server by searching for MCP: Add Server and entering the connection details above.

The Connext MCP server currently offers several tools: ask_connext_question and XML, Python, and C++11 code validation tools.

The ask_connext_question tool allows your agent of choice to send questions to Connext AI and use the response as context for future interactions. For example, an agent could ask a question about how to configure a specific QoS policy in Connext, and then use the response to inform its next steps in configuring a system design.

The code validation tools allow your agent to submit a snippet of code for validation (XML-schema validation, mypy static type validation, or C++11 symbol validation).

Prompt examples

When interacting with Connext AI, the more specific your prompt, the better the response. Here are key elements to include for an effective prompt:

1. Product Mention: Specify the product you're inquiring about. For example, specify Routing Service or Micro in the prompt.
2. Specific Request: Clearly state your question or the information you need. For example, "help me understand how endpoint discovery works," or "write a modern C++ code example to create a content filtered topic."
3. Context/Details: Provide additional context or details to help the chatbot better understand your request. For example, note your specific use case, such as "for a surgical robotic application".

The following are detailed example prompts to get you started:

• Can you explain how to configure QoS policies in Connext for optimizing reliability in a teleoperated robotic application?
• How can I reduce latency in my Connext DDS system for a real-time distributed simulation application?
• How does Connext Micro differ from Connext Professional for IoT device integration?

Accessing Interaction History

Connext AI maintains a history of your interactions, allowing you to review past conversations. This feature is particularly useful for tracking the progression of inquiries or revisiting previous responses.

In the Connext Chatbot, your interaction history is displayed to the left of the chat panel; click the Show icon to open a previous chat session. After opening an older session, you can continue to ask follow up questions within the same context. To remove a session from your history, click the Remove icon.

Note that conversations are still stored in the backend and will not be removed unless you explicitly request it by contacting privacy@rti.com.

Providing Feedback

To continually improve the quality of its responses, the Connext Chatbot includes a feedback mechanism. To help identify areas for improvement, you can rate the responses using the rating scale displayed at the end of each reply. Additionally, you can provide more detailed feedback via email to connext-ai-feedback@rti.com.