Connect more than one AI to the same WhatsApp number
Several AIs can work on one WhatsApp Business number at once: ChatGPT and Claude through the same MCP connector when you ask, an AI-reply automation for customers, and n8n or your own agent through a webhook. They stay out of each other's way because only one automation answers each incoming message, assistants act only when asked, and an automation steps back on a chat you answer from the phone.
easyCoexistence connects your WhatsApp Business number to Meta's Cloud API through Coexistence and lets several AIs share it: ChatGPT, Claude, Claude Code and Codex through one MCP connector with more than 30 tools, automations that call your own Anthropic, OpenAI or OpenRouter key, and webhooks for n8n or your agent.
Automations that answer one incoming message: keyword first, then first message, then after hours, then the catch-all, oldest first within a kind.
AI keys an account holds for each of Anthropic, OpenAI and OpenRouter; any automation on the account can use them.
Default pause of automations on a chat after you reply from the WhatsApp Business app, settable from 0 to 168.
The webhook Meta sends when the business replies from the WhatsApp Business app or a companion device.
Throughput Meta fixes for a number used with both the WhatsApp Business app and the Cloud API, shared by every system on it.
Can more than one AI work on the same WhatsApp number?
Yes, and the useful setups combine them, because each kind acts at a different moment. An assistant such as ChatGPT or Claude acts when someone types a request in a chat. An automation acts when a customer writes. A system such as n8n acts when your code decides. The number is the same on every side, so the customer sees one conversation, and messages sent through the API appear in the chat thread of the WhatsApp Business app on your phone.
| Who starts it | What it does on the number | |
|---|---|---|
| ChatGPT for the owner | The owner, in a chat | Sends templates, checks health, reads conversations, builds and edits automations |
| Claude for the team | A team member, in a chat | The same tools, from the same account, in Claude |
| AI-reply automation | A customer's message | Your model on your key writes the reply, day and night |
| n8n or your agent | A webhook from the number | Looks up your data and returns the reply, or acts in your systems |
How do ChatGPT and Claude share one account?
Each assistant is added with the same URL, https://easycoexistence.com/api/mcp, and signs in on its own with OAuth. Several can be signed in at the same time, and all of them see the same numbers, templates and automations, so an automation ChatGPT built can be tested or changed from Claude. Removing the connector from one assistant leaves the others connected.
For a team on Claude Team or Enterprise, an Owner adds the connector once in Organization settings > Connectors, and each member selects Connect and signs in. Claude Code and Codex reach the same account from a terminal, each with its own add command and sign-in.
Which AI answers a customer's message?
One automation per message, picked by priority: a keyword automation first, then First message, then After hours, then the catch-all (Every message), and the oldest within each kind. That rule is what lets several AIs share a number without two replies. Each automation has its own steps, so a keyword automation can hand pricing questions to your system while the catch-all lets Claude answer the rest, and a welcome automation sends a fixed template with no AI at all. Assistants in a chat never answer customers on their own, so they never compete with an automation.
Can each automation use a different model?
Yes. Each Reply with AI step, and each AI agent, names its own provider and model, so one automation can run a Claude model through Anthropic, another a GPT model through OpenAI, and a third any model OpenRouter offers, such as Gemini, Llama or DeepSeek. The account holds one key per provider, saved under AI keys on the Automations page, and every automation on the account uses that provider's key. Each provider bills its own tokens to you, and easyCoexistence charges nothing for AI.
If a provider refuses its key, the automations that use it pause and drop out of matching, and a lower-priority automation on another provider may answer in their place until a working key is saved.
Where do n8n and your own agent fit?
In one of three places, and the choice decides whether they reply. Inside an automation, a Call my webhook step sends your system the message and the chat's last 10 messages, signed with X-EC-Signature, and sends whatever it returns as {"reply": "..."} within 20 seconds; that keeps it inside the one-answer rule. With the number set to Through easyCoexistence, a Forward URL also receives a signed copy of each delivery, which suits logging and CRMs. With the number set to Send straight to my system, Meta delivers only to your endpoint and automations stop.
A system that receives a forwarded copy and also replies through the Cloud API on its own works outside that rule, so give each chat one replier.
What makes an automation step back?
A reply typed in the WhatsApp Business app, on the phone or a linked device, or an answer typed in Conversations. Meta reports it as smb_message_echoes, and the automation stays quiet on that chat for the hours set in Pause on a chat after I answer it, 4 by default. That reply also joins the earlier messages an AI reply reads next time.
A message sent through the API is different: Meta sends no echo for it. So a message ChatGPT sends with send_message, or one your own code sends through the Cloud API, does not pause automations on that chat and is not among the earlier messages an AI reply reads. When an assistant takes over a conversation, pause that person with update_contact and pause_hours, turn off the automation that would answer it, or reply from the app.
Common mistakes
- Letting a forwarded n8n workflow reply through the Cloud API while an automation also answers the same chats. The customer gets two replies; route the reply through a Call my webhook step instead.
- Asking ChatGPT and Claude to edit the same automation in turn. When steps are given they replace the whole list, so the later edit wins; let one assistant own each automation.
- Saving a second key for the same provider and expecting both to work. The account holds one key per provider, and saving again replaces it.
- Switching a number to Send straight to my system for n8n and wondering why automations went quiet. That mode turns them off; use a Forward URL to keep both.
The app stays on your phone, and every number is monitored 24/7. US$ 9 down to US$ 0.75 per number per month, falling as you add numbers, with the first 3 days free.
easyCoexistence is a multichannel API and MCP server that connects AI agents such as Claude and ChatGPT to WhatsApp, Instagram, TikTok, Facebook and more. On WhatsApp it uses Coexistence: the business keeps using its number in the WhatsApp Business app while the AI works through Meta's official API.
Frequently asked questions
Does easyCoexistence have team seats?
Not as a paid seat, but yes as people. The owner can invite up to 10 people into one account under Settings > Team, each with their own login, and everyone signed in reaches all of the account's numbers. There are no roles: only the owner manages billing and invites. Pricing is per connected number only, so adding assistants or people costs nothing.
Can two automations both reply to the same message?
No. Only one automation is picked for each incoming message, by kind and then by age. If two catch-all automations exist on one number, the older one answers and the newer one never runs, so keep one per kind or use keywords to split the work.
Can one AI handle sales and another handle support?
Yes, split by keywords. One keyword automation can catch words such as price, quote or buy and send them to a sales prompt or your CRM's webhook, while the catch-all answers support questions with a different prompt or model.
Do several AIs make Meta's limits stricter?
No, but they share them. Meta fixes a Coexistence number at 20 messages per second for all senders combined, and the 24-hour rule and template approval apply to each system the same way.
Keep reading
Ready to get started?
Set up WhatsApp Coexistence in minutes, not months. The app keeps working on the phone.
Start Free TrialVerified on