Skip to content

Connect your AI (MCP)

Behaview works with your AI assistant. Using the Model Context Protocol (MCP), you can connect AI tools like Claude, ChatGPT, and Cursor directly to your Behaview organization so they can do real work, under your account’s permissions, acting as you. It is one of the fastest ways to set up and run Behaview, so set it up early.

Once connected, an AI client can do real supervision work in Behaview, all through conversation. Some examples of what you can ask:

Capture a note

“Take a note on Jordan to review prompt fading at our next supervision.”

Record an observation

“Add an observation for Jordan on discrete-trial teaching: clean trials today, and prompts were faded on time.”

Create skills

“Create a set of skills for our clinic’s toilet-training protocol that I can assign to our RBTs.”

Prep a supervision session

“Prep me for tomorrow’s supervision with Jordan: which RBT Test Content Outline items should we cover?”

Draft a review

“Based on my notes, observations, and Session Plans with Jordan, draft a performance review.”

Behind these sit all of Behaview’s building blocks: skills, Session Plans and observations, notes, dashboards, and more. Your AI calls the two people on a Session Plan the Observer and the Observee, as Behaview does.

Behaview’s MCP server also exposes guided workflows: step-by-step instructions for common jobs, such as setting up a Session Plan, recording skill observations, writing skills, or working through your own skills with AI coaching. Your AI follows behaviorally sound practice rather than guessing, so it acts as a real assistant to you as a supervisor, not just a way to enter data.

Behaview’s MCP server URL has the form https://manage.<region>.behaview.net/mcp, where <region> is your organization’s two-letter region code. Use:

  • Canada (ca): https://manage.ca.behaview.net/mcp
  • United States (us): https://manage.us.behaview.net/mcp

You can tell your region from the address bar when you’re signed in to Behaview.

Add that URL as a remote (custom) MCP server in your AI tool, then sign in when prompted. Behaview uses a secure sign-in (OAuth), so the tool acts as you with your access. No credentials are ever copied or shared. The connection stays signed in when you sign out of Behaview, and ends if you don’t use it for 90 days. To disconnect, remove Behaview from your AI tool.

The exact steps differ slightly by tool:

  • Claude: add Behaview as a custom connector (Settings → Connectors), then paste the MCP URL. See Anthropic’s connector help.
  • ChatGPT: open Settings → Plugins and choose Add MCP Server. Enter Behaview as the name. Select Streamable HTTP, paste the MCP URL into the server URL field, then save the server. If ChatGPT shows Restart, select it to apply the change. Select Authenticate if shown, then sign in when prompted and approve access.
  • Cursor and other MCP-capable tools: add the URL as an MCP server in the tool’s settings.

Once connected, walk through Your first week: your AI can do most of the setup with you.

  • Sign-in says your username or password is incorrect, or your AI finds only a new, empty organization. The region in the MCP URL must match your organization’s. Check the address you sign in to Behaview at, then reconnect with that region’s URL.
  • Your AI tool asks you to sign in again. A connection ends after 90 days without use, or when you no longer belong to any Behaview organization. Sign in again to reconnect. If you were removed from your organization, an admin must add you back first.
  • Your AI works in the wrong organization. If you belong to more than one, name the organization in your request.
  • Your AI says it can’t do something. It can do only what your role allows in Behaview, and a few tasks can be done only in the app. Ask an admin if you need more access.

A connected AI tool can read and change your organization’s data with your permissions, including deleting records you are allowed to delete. The data it reads is sent to the tool’s provider and handled under that provider’s terms, so check your AI tool’s data settings. Behaview logs which Behaview action each request used and which app sent it, but not your conversation with your AI tool.