# A Few Seconds to Reconsider Before a Message Sends Chat is instant by design, which is exactly the problem when the other side of the conversation is an agent that starts acting the moment your message lands. Hit enter on a typo, an incomplete thought, or a command you didn't mean to send yet, and there's no getting it back — the agent is already reading it. Message Send Delay adds a brief buffer between hitting send and the message actually reaching the agent, with a chance to change your mind before it does.  ## A Countdown, Not a Guessing Game Turn it on from **Workspace Settings → General → Message Send Delay** and pick how long a buffer you want: 3, 5, 10, 15, 30, or 60 seconds. It's a per-workspace setting, so a workspace where an agent moves fast on production infra can run a longer delay than one where you're just brainstorming. Send a message with a delay active, and the composer swaps to a countdown: **"Sending in Ns…"**, alongside two buttons.  **Send Now** delivers immediately, skipping the rest of the countdown — useful once you've re-read the message and you're sure. **Cancel** does the opposite: it pulls the message back out, unsent, and puts the exact text and any attachments right back in the input box, ready to edit. Nothing reaches the agent, and nothing is lost. Leave it alone, and the countdown does what a normal chat message would do anyway — it delivers on its own once the timer hits zero. The delay only ever adds a pause; it never silently drops a message. ## Off by Default Every existing and new workspace ships with the delay off (0 seconds), so nothing about how messages are sent changes unless you turn it on. This is a setting you opt into for the workspaces where a moment's pause is worth the wait — not a behavior change imposed on every conversation. ## Why This Matters An agent doesn't pause to double-check with you before it starts on whatever you just typed. That immediacy is the point of the tool, but it means a fumbled message costs more than it would in a conversation between two people. Message Send Delay puts a small, optional gap back into that immediacy — long enough to catch a typo or reconsider, short enough not to get in the way when you don't need it. *AgentRQ is currently in public beta. Join our [GitHub community](https://github.com/agentrq/agentrq) to help shape the future of human-agent collaboration.*