Bedones plugs into ChatGPT and Claude through a single connector, at the address https://api-moderator.bedones.com/mcp. Once that connector is added, you read and answer your WhatsApp, Facebook, Instagram and TikTok messages and comments without leaving your chat with the AI. Setup takes five steps and ends on an authorisation screen where you pick the organisation the AI is allowed to work on.

What ChatGPT and Claude do once Bedones is connected

Once Bedones is connected, ChatGPT and Claude act on your real accounts: they list your unread conversations, draft and send a reply, hide an unwanted comment, pull a product out of your catalogue or open a ticket. You type the request in plain English, in the chat window, and the AI carries it out inside Bedones.

The requests below work as soon as setup finishes, with nothing else to configure on your side. They are written the way you would say them to a human assistant: no keywords, no special syntax, no technical tool names to remember. The AI picks the matching Bedones action by itself.

  • “Show me my unread conversations across every channel”
  • “Reply to Marie’s last WhatsApp message with our opening hours”
  • “Hide the spam comments on my latest Facebook post”
  • “Find the red dress in the catalogue and send it in this conversation”
  • “Open a priority ticket for this refund request”

One connector covers all four networks at once, so you never plug WhatsApp in on one side and TikTok on the other. The accounts already linked to your Bedones organisation become visible to the AI at the moment you authorise it, and the full list of exposed tools is described in how the Bedones MCP connector works.

What you need before adding the connector

Three things are enough to add the Bedones connector, and none of them calls for technical skill. The first is a Bedones account with at least one network already linked, since the AI can only read conversations that exist. The other two are access to custom connectors in your AI, and the address of the Bedones server.

  • A Bedones account with at least one network connected: WhatsApp, Facebook, Instagram or TikTok
  • A ChatGPT account with developer mode and connectors enabled, or a Claude Pro, Team or Enterprise account
  • The connector address: https://api-moderator.bedones.com/mcp

The two AIs do not expose custom connectors in the same place. Claude offers them on the Pro, Team and Enterprise plans. ChatGPT shows them once developer mode is switched on in settings: without that, the manual add option never appears, even though the Connectors menu already sits in your account.

Bedones does not have to appear in the public app directory of ChatGPT or Claude. Depending on your country, your plan and the date, the app may not be listed there yet. Adding it by hand as a custom connector gives exactly the same result: same address, same authorisation, same capabilities afterwards.

Adding the Bedones connector in Claude

In Claude, the Bedones connector is added from Settings, then Connectors, using the Add custom connector button. You paste the address and confirm: Claude reads the server configuration on its own and opens the secure connection, with no other field to fill in on your part.

  1. Open Claude, then Settings, then Connectors
  2. Click Add custom connector
  3. Paste the address https://api-moderator.bedones.com/mcp and confirm
  4. On the Bedones “Authorise AI access” screen, sign in, choose your organisation and click Authorise
  5. Test it with “List my connected social accounts”

The authorisation screen belongs to Bedones rather than to Claude, and that is where the scope of the access is settled. The organisation selected at that moment is the only one the AI will be able to work on afterwards. If you run several of them, check which one is ticked before clicking Authorise.

Step five is there as a check. A simple read request confirms in one go that the connector answers, that the authorisation went through, and that the AI is looking at the right organisation. If the list comes back empty, no network is linked in Bedones yet, and that is where to pick things up.

Adding the Bedones connector in ChatGPT

In ChatGPT, the Bedones connector is added from Settings, then Connectors, with developer mode on, by choosing Add custom connector and then the Server URL option. The authentication type to select is OAuth: that is what triggers the Bedones authorisation screen instead of a request for a key.

  1. Open ChatGPT, then Settings, then Connectors, with developer mode enabled
  2. Choose Add custom connector, then Server URL
  3. Paste https://api-moderator.bedones.com/mcp and select OAuth authentication
  4. Click Create, authorise the access and select your organisation on the Bedones screen
  5. Test it with “Which messages are waiting for a reply?”

The ChatGPT procedure differs from Claude’s on two points only. Developer mode has to be enabled beforehand in settings, and the authentication type is chosen explicitly instead of being detected. Everything else is identical, right through to the Bedones screen and the choice of organisation.

The final test doubles as a first demonstration: the question forces the AI to go through your conversations and sort out what is still waiting. That is already the move you will repeat every morning. For an agent tuned to one specific market, see how an agent is configured for Cameroon.

What the AI sees, and what it never will

The AI sees what the Bedones organisation you selected contains, and nothing else. It never gets your Bedones password, nor the access tokens for your Meta and TikTok accounts: those tokens do not leave Bedones. You are the one who signs in, on a Bedones screen, and the AI simply uses the permission you granted it there.

The connection runs on OAuth 2.0, the standard sitting behind a “Sign in with Google” button. Every tool Bedones exposes is labelled according to whether it reads or changes something, destructive actions stay restricted to administrator accounts, and the AI only acts on a request made in the conversation.

Removing the access is as easy as granting it: delete the connector from your ChatGPT or Claude settings and the authorisation drops, without touching the social accounts linked in Bedones. Nothing you configured in Bedones is lost, and you can authorise the connector again later. With the connection in place, the next question is what to let the agent handle on its own.