AI TOOLS FOR BUSY BUSINESS OWNERS & CREATORS.

I Built a Custom Chat Interface for OpenClaw

> TLDR

$ problem: discord shows thinking blocks in every response

$ solution: built a custom chat interface

$ time: fred built it in a few hours

$ cost: $0 (runs on your own machine)

I was getting super annoyed that Discord shows the "thinking" blocks in every response. Couldn't figure out how to turn it off, so I just had Fred (my OpenClaw agent) build me a new interface.

Now it's clean like Claude Code. I still see the "thinking" in real time, but then it delivers just answers.

Still building out features, but it works great so far.

Fred Chat interface showing a conversation with the AI agent
DISCORD (THE PROBLEM)
thinking blocks shown in every response
no gateway controls
no voice chat with your bot
no way to install as standalone app
no token tracking
no auto-compact
FRED CHAT (THE FIX)
clean answers, thinking hidden
1-click gateway restart
voice chat built in
PWA installable on mobile + desktop
live token counter
auto-compact at 150k tokens

What It Does (Beyond Stock Discord)

Here's everything the custom interface has that Discord doesn't:

feature Discord Fred Chat
filters thinking blocks [N] [Y]
restart gateway in 1 click [N] [Y]
compact or new chat in 1 click [N] [Y]
voice chat (real-time back and forth) [N] [Y]
install as standalone app (mobile/desktop) [N] [Y]
shows gateway connection status [N] [Y]
shows token usage [N] [Y]
auto-compact at 150k tokens [N] [Y]
scheduled messages [N] [Y]
cron routing [N] [Y]

How It Actually Works

The interface is stupid simple. It's just a web page that talks to the OpenClaw gateway through WebSockets. Fred built the whole thing — I didn't write a single line of code.

TECH STACK
server: Node.js
connection: WebSocket to OpenClaw gateway
voice (TTS): Edge TTS (Microsoft)
voice (STT): OpenAI Whisper API
mobile install: PWA manifest
cost: $0/mo (runs locally) + $5.40/mo for voice at 1hr/day
// real talk

The chances of me being able to do any of this without my bot are zero. I'm not technical. I've never written a line of code. I didn't even know what the terminal on my Mac was until a few months ago.

I just told Fred what I wanted and he built it. Took a few hours. That's it.

Voice Chat Is the Best Part

I even added voice chat so I can talk to Fred right in the browser. Same speech-to-text, same voice back. Real-time conversation. No waiting.

It's the same tech I wrote about in my other post about phone calls. Works perfectly on mobile too — I installed it as a standalone app on my iPhone and now I just tap the icon and start talking.

// voice costs

Edge TTS is free (Microsoft's text-to-speech). OpenAI's gpt-4o-mini-transcribe is $0.003/min. If you talk to your AI for 1 hour per day, that's $5.40/mo. Unlimited voice conversations for the price of a coffee.

How It's Wired Up

BROWSER UI (chat window) | | WebSocket | Node.js Server | |──> OpenClaw Gateway (agents, tools, responses) | |──> Edge TTS (free text-to-speech) | └──> OpenAI gpt-4o-mini-transcribe (speech-to-text, $0.003/min)

Why Bother?

Discord is fine for most people. But I use this thing all day, every day. The thinking blocks were driving me crazy. I wanted something cleaner.

And now I have it. No waiting on Discord to add features. No dealing with their UI decisions. If I want something, Fred just adds it.

That's the whole point of running your own AI stack. You're not stuck with what some company decides to give you.

<< BACK TO POSTS