TimelinesAI ChatGPT Agents let you put OpenAI's AI models to work on your WhatsApp numbers. You create and configure each Agent inside TimelinesAI — its name, instructions, and model — powered by your own OpenAI API key. Once set up, an Agent automatically replies to incoming WhatsApp messages on your behalf.
Note: As of August 20, 2026, TimelinesAI ChatGPT Agents run on OpenAI's current Responses API, replacing the deprecated Assistants API. See Deprecated OpenAI Assistants below to read more.
Key facts
You can assign an agent to a specific WhatsApp number or to "All Numbers" (covers all current and future numbers in your workspace).
Only one active agent per WhatsApp number is allowed.
In direct chats, the agent replies automatically. In group chats, an agent is never activated automatically and must be enabled manually.
Attachments are ignored — the agent replies to text messages.
The agent automatically disables itself in a chat as soon as an outgoing message is sent by anyone other than the agent — a human teammate or any automation (Workflow Builder, Zapier/Make, the Public API, or a CRM). Once disabled, it won't re-join the chat — even for new incoming messages — until it's re-enabled.
Requirements
An OpenAI account with an API key and sufficient token balance.
A TimelinesAI workspace with available automation credits.
Setup
Add your OpenAI API key
Go to Automations → ChatGPT Agents (https://app.timelines.ai/openai/assistants/settings/) and add your OpenAI API key. You can get a key from platform.openai.com/api-keys.
Connect an Agent
Click "Connect an Agent" and follow the wizard.
Step 1 — Choose where the agent works
WhatsApp Account/s to work with an AI agent — pick a specific number or All Numbers.
Choose Only in new direct chats or In any direct chat and existing group chats.
Note: When a new text message arrives in a chat that matches your settings, the AI agent replies automatically. Attachments are ignored. For automatic replies to work, the agent must be active in the chat — in group chats, agents are never activated automatically and must be enabled manually.
Step 2 — Select or create the AI agent
Choose an existing agent from Select AI agent, or click Add new agent to create one. When creating a new agent, fill in:
Agent name — a name for your agent.
Instructions — tell the agent how to reply (for example, "Reply clearly and briefly."). TimelinesAI uses these instructions, together with the chat history it sends, on every response.
Advanced → Model — the OpenAI model the agent uses (default gpt-4.1-mini).
Advanced → Response format — Text (default)
Returning user? If you created deprecated assistants in OpenAI and had them active in TimelinesAI before August 20, 2026, they were converted to Agents automatically and will appear in the Select AI agent list. See Deprecated OpenAI Assistants below.
Step 3 — Name and activate
Auto response override (seconds) — how long the agent waits after the first new message before replying, batching any messages received during that window into one reply (the countdown does not reset with each new message). Default 5 seconds.
Display name — a friendly name shown in TimelinesAI. Keep the existing name from the OpenAI platform, or enter a custom one.
Choose Save to save without turning it on, or Save and Activate to start replying on the selected WhatsApp number(s).
Deprecated OpenAI Assistants
If you created assistants in OpenAI and connected them to TimelinesAI before August 20, 2026, you don't need to do anything — your actively-used assistants were converted to Agents automatically and keep working:
OpenAI is deprecating the Assistants API, and assistants must be moved to OpenAI's new Responses API. We handled this conversion for you, following OpenAI's official Assistants migration guide.
Your converted Agents stay connected to the same WhatsApp numbers as before, so replies keep flowing without interruption. They appear in the Select AI agent list when you connect an agent.
Any new agents you create in TimelinesAI now also run on the Responses API.
Important: We converted only the assistants that were in active use. If you have other assistants in your OpenAI account that you'd like to keep, back them up before OpenAI removes the Assistants API.
Have feedback about agents running on the new Responses API? Book a meeting to discuss your feedback.
Multiple WhatsApp numbers
Each seat (user profile) in TimelinesAI connects to one WhatsApp number.
To connect 3 numbers, you need 3 seats.
You can then assign one ChatGPT agent to "All Numbers," so it responds across all connected numbers automatically.
Pricing
Each agent reply uses 1 TimelinesAI automation credit + OpenAI tokens (billed by OpenAI based on your usage).
Re-enabling an agent
If an agent was disabled in a chat (manually, or automatically after an outgoing message), turn it back on from the chat's details sidebar using the ChatGPT Agent toggle ("enable for this chat"). After re-enabling, the agent only responds to messages received after it was re-enabled — not to messages that arrived while it was off.
Why did my agent stop replying?
The most common reason is the automatic disable described above — for example, a follow-up message sent by an automation in that chat. You can confirm it in the chat itself: the chat's event log shows "ChatGPT Agent auto-disabled due to outgoing message from human agent" at the moment it happened. This appears in the chat's in-app event log; it does not appear in the agent Activity Log.
Troubleshooting
Check the Activity Log for agent performance details. If an agent shows a red "Fix issues" flag, it usually means the OpenAI API key is invalid or has insufficient balance.
In case of any questions, contact our dedicated customer support team.