Beta — limited early access

Deploy AI agents.Without DevOps.

Connect to Slack, WhatsApp & Telegram in minutes — no servers, no YAML, no infrastructure teams.

ConnectConfigureDeploy

No credit card required · Free to start

200+ teams already building with ClawDuck

Telegram@clawduck_bot
Connected
0:04
Clawduck Bot

Features

Everything you need to run AI agents in production.

Multi-model

Claude, GPT, Gemini, Grok, Llama, and more - model routing and credits managed for you out of the box.

Channels first

Wire Slack, Discord, Telegram, or WhatsApp with guided setup flows and health checks.

AgentOS custom agents

Build specialized agents with your own instructions, tools, API hooks, and MCP servers in minutes.

AgentOS Harness

Orchestrate teams, workflows, traces, and memory with a production runtime - paid tiers only (Starter, Pro, Team).

Local voice STT

Send voice messages on Telegram - transcribed on-device. No third-party API, no audio leaves your pod.

Local memory

On-device vector embedding power long-term memory with semantic search - no external embedding API required.

Enterprise-grade logic & observability

Everything you need to ship confident agents.

Webhook Trigger
LLM: Claude 3
Tool: Search
Slack Message Out
Writer
Researcher
Review
company_guidelines.pdf
When did we launch the V2 API?
Parsing history... you launched V2 on Oct 14th.
Can you generate a summary report?
Sure, here is your summary from those logs.

Tutorial

See how to go from an OpenClaw or ZeroClaw config to a live, 24/7 instance in just a couple of minutes. No cloud consoles, clusters, or YAML spelunking required.

Prefer to watch on YouTube? Open the video in a new tab.

Voice & memory - fully local

Your pod turns voice messages into text and remembers past conversations for you. No audio or text is ever sent to outside services - everything stays inside your own private container.

  • Send a voice note in Telegram - your agent reads it, just like text
  • Quickly search across past chats like a personal memory
  • No extra accounts or surprise bills - everything runs on your pod
  • Optionally connect to cloud services if you prefer
Voice pipeline (runs inside your pod)
Voice message from chat
recorded inside your pod
Text
Private speech-to-text service
runs next to your agent, not in the cloud
LLM
Agent processes transcript
with full conversation + memory context
Reply sent back in Telegram
zero audio data leaves your pod

API key access

Pro and Team plans include API keys to interact with your agent gateway programmatically. Send messages, invoke tools, and build custom integrations - scripts, apps, or webhooks.

After deploying on Pro or Team, create your key under Dashboard → API for your OpenClaw agent.

curl -X POST https://clawduck.com/api/gateway/invoke \
  -H "Authorization: Bearer cd_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "Hello, what can you help me with?"}'

Pricing

From free to team - multi-agent personas, channel routing, and up to 3 agents for teams.

Free

Free
  • $1 token credits/mo
  • 1 agent
  • Open-source models
  • AI buddies (personas & channel routing)
  • 24/7 always-on instance
Get Started Free

Starter

$9/mo
  • $5 token credits/mo
  • 1 agent
  • Open-source models
  • AI buddies (personas & channel routing)
  • 24/7 always-on instance
  • AgentOS Harness access
  • Local voice STT & memory
Subscribe - $9/mo
Recommended

Pro

$29/mo
  • $20 token credits/mo
  • 1 agent
  • All models incl. Claude & GPT-5
  • AI buddies (personas & channel routing)
  • 24/7 always-on instance
  • AgentOS Harness access
  • Local voice STT & memory
  • OpenClaw gateway API (Dashboard → API)
Subscribe - $29/mo

Team

$79/mo
  • $40 token credits/mo
  • 3 agents with buddies each
  • All models incl. Claude & GPT-5
  • AI buddies (personas & channel routing)
  • 24/7 always-on instance
  • AgentOS Harness access
  • Local voice STT & memory
  • OpenClaw gateway API (Dashboard → API)
Subscribe - $79/mo

How it works

ClawDuck wraps your preferred LLM and messaging channel in a production‑grade runtime, so you can go from YAML to running agent in minutes instead of weeks.

1. Choose your agent & model

Pick OpenClaw or ZeroClaw, then choose the brain (Kimi, Grok, Gemini, Llama) and where it lives.

2. Connect your keys

Securely attach API keys and channel credentials—no secret sprinkling across random scripts.

3. Deploy to ClawDuck

We spin up a dedicated, always‑on instance with health checks, autoscaling, and logging.

4. Configure buddies

Add AI buddies, route channels to personas, and roll out changes safely without downtime.