Do I need to connect Telegram before I can use my bot?

No. Quick Chat opens a conversation from the instance card in the dashboard, so you can talk to a bot the moment it is running. Connecting a platform is for reaching it from your phone or giving other people access.

Does Quick Chat remember the conversation?

On OpenClaw, yes — the conversation is stored per user and per instance, so it is still there when you return, and any separated reasoning appears in a collapsible block that is not kept in the transcript. Hermes can resume a conversation where its gateway session is available, but otherwise uses a stateless path whose transcript lives only in the open page. Replies stream in as the model writes them on both.

Does it work on both OpenClaw and Hermes?

Yes, from the instance card on either framework. The bot must be running. On Hermes it additionally requires an active subscription or a running free trial, and an API server port, which is allocated when the instance is created rather than by the image it runs.

← Home

Guide

Quick Chat

You deploy a bot and the obvious next step looks like a chore: go and register a Telegram bot, or set up Discord, before you can say hello. Quick Chat removes that step. It is a button on the instance card that opens a conversation right there in the dashboard. Works on OpenClaw and Hermes.

Where It Is

Open the dashboard and find the card for a running bot. Quick Chat sits in the row of buttons on the card, next to Terminal. Click it and the conversation opens in place — there is nothing to install and nothing to configure.

The bot has to be running. The panel is only offered for a running instance, and it closes on its own if the bot stops while you have it open, which is expected rather than a crash.

What You Get

A real conversation, not a one-shot box. On OpenClaw the chat keeps its history: your messages and the bot’s replies are stored against that instance, so closing the dashboard and coming back tomorrow continues where you left off rather than starting cold.
Streaming replies. Text appears as the model writes it, so a long answer is readable while it is still being produced.
A collapsible thinking block, on OpenClaw. When the model separates its reasoning from its answer, the reasoning is shown in a block you can expand. It is deliberately not kept in the transcript — the answer is what persists.

What It Is Good For

Proving the bot works. Straight after a deploy, one message confirms the model is reachable and answering. If something is misconfigured you find out in ten seconds instead of after twenty minutes of Telegram setup.
Testing a change. Switched model, installed a tool, edited a skill? Ask something that exercises it and see the result immediately.
Actual work. Plenty of people never connect a platform at all. If you are at your desk with the dashboard open, this is simply where you talk to your bot.

What It Is Not

Quick Chat lives in the dashboard, which is its strength and its limit. It is not a way to reach your bot from your phone on the train, and it is not how you give other people access. For either of those, connect a real channel — Telegram on OpenClaw, Telegram on Hermes, or Discord.

It is also distinct from the bot’s own web interface. On OpenClaw, Open Gateway takes you to the gateway UI; on Hermes, Open Dashboard takes you to the Hermes dashboard. Those are the full applications. Quick Chat is the shortcut.

OpenClaw and Hermes

The button is on both cards and behaves the same way. Two differences worth knowing:

Subscription. On Hermes, Quick Chat needs an active subscription or a running free trial; outside those it prompts you to subscribe rather than opening. OpenClaw’s Quick Chat has no such check.
History. OpenClaw stores the conversation server-side, so it is there when you come back. Hermes can resume a conversation where the bot’s own gateway session is available, but it otherwise talks over a stateless path whose transcript lives in the open page — close the tab and that one is gone. Clearing a Hermes chat is likewise a local action.
The button itself. On Hermes it depends on the instance having an API server port, which is allocated when the instance is created. That is a property of when the bot was made, not of the image it runs — so upgrading an older bot does not make the button appear.

Troubleshooting

No Quick Chat button. Check the bot is running — the whole button row is for running instances. On Hermes, also check the subscription or trial state.
The panel closed by itself. The instance stopped. Start it and reopen — on OpenClaw the conversation is still there; on the Hermes stateless path the visible transcript starts fresh.
An error about credits or billing. That is the model provider refusing the request, not the chat. Check the model and key on the card — see BYOK on OpenClaw or BYOK on Hermes.
Replies are slow on the first message. A bot that has just started is warming up. The second message is the honest measurement.

Try It

Open a running bot and send it one sentence about something it should know how to do. That single exchange tells you whether the model, the key and the configuration are all working — which is the thing worth knowing before you invest in connecting a platform.