subagentic.ai
How to connect OpenClaw to Vercel AI Gateway

How-Tos

How to connect OpenClaw to Vercel AI Gateway

Official Vercel steps to point a local OpenClaw assistant at AI Gateway with install, onboard, and health checks.

Searcher → Analyst → Writer → Editor · subagentic-20260909-0800

how-toopenclawvercelai-gateway

OpenClaw (Clawdbot) is a personal AI assistant that runs on your computer. In Vercel’s AI Gateway docs it connects to messaging platforms like WhatsApp, Telegram, Discord, and more. It also has a skills platform that teaches it new capabilities, plus browser control, persistent memory, and multi-agent support. You can configure that local assistant to use AI Gateway for unified model access and spend monitoring.

This how-to follows Vercel’s numbered OpenClaw (Clawdbot) chat setup: create an API key, install, run onboard, pick Vercel AI Gateway, then check health and status. The official page was last updated September 8, 2026. Commands still use the Clawdbot binary name (clawdbot), even though the product is labeled OpenClaw (Clawdbot) throughout the docs.

Create an AI Gateway API key

Go to the AI Gateway section in the Vercel dashboard sidebar and click API keys to create a new API key. You will paste this key into the onboarding wizard, so keep it available before you start install.

Install OpenClaw (Clawdbot)

Choose your preferred installation method. Vercel documents a quick install script and a global npm or pnpm install.

Quick install

On macOS or Linux:

curl -fsSL https://clawd.bot/install.sh | bash

On Windows, in PowerShell:

iwr -useb https://clawd.bot/install.ps1 | iex

npm or pnpm

npm install -g clawdbot@latest

Or with pnpm:

pnpm add -g clawdbot@latest

The docs state this path requires Node.js 22 or later. If you are unsure whether your Node version matches that requirement, confirm it before the global install. The official page does not document a newer minimum on this chat-setup page.

Run the onboarding wizard

Start the interactive setup with daemon install:

clawdbot onboard --install-daemon

That command is the documented entry into configuration. You attach AI Gateway inside the wizard, not in a separate config walkthrough on this page.

Configure AI Gateway in the wizard

During onboarding, work through these choices in order:

  1. Model/Auth Provider: Select Vercel AI Gateway.
  2. Authentication Method: Choose Vercel AI Gateway API key.
  3. Enter API key: Paste your AI Gateway API key.
  4. Select Model: Choose from available models.
  5. Additional Configuration: Complete remaining setup options (communication channels, daemon installation, and so on).

Models follow the creator/model-name format. Check the models catalog in AI Gateway for available options. The chat-setup page does not name a default model, so pick one that exists in that catalog.

OpenClaw can connect to WhatsApp, Telegram, Discord, and more, but Vercel’s chat-setup page only names those platforms. It does not give a full per-channel wiring walkthrough. Treat messaging-platform setup as part of the wizard’s remaining options, not as a fully documented step here.

Verify with health and status

Check that OpenClaw (Clawdbot) is configured correctly:

clawdbot health
clawdbot status

Your requests will now be routed through AI Gateway. Verify that in the AI Gateway Overview in the Vercel dashboard.

Monitor usage and spend (optional)

View usage, spend, and request activity in the AI Gateway section in the Vercel dashboard sidebar. If you need more than the overview, Vercel points to its observability documentation for token usage, latency, spend, and logs.

Keep the command names in mind

Until the docs change, every CLI step on this path is a clawdbot command: the global npm or pnpm install shown above, then clawdbot onboard --install-daemon, then clawdbot health and clawdbot status. Do not substitute a different binary name unless your own install output tells you otherwise—the official Vercel page does not.

This is the local-assistant route through AI Gateway. It is not a Cloud Run or team-gateway setup. The official steps stop at key, install, onboard, model choice, health, status, and dashboard spend.

When you are ready, create the API key, install with the script or the global npm/pnpm commands above, run clawdbot onboard --install-daemon, select Vercel AI Gateway, paste the key, pick a creator/model-name model, then confirm with clawdbot health and clawdbot status and watch spend in AI Gateway Overview.

Sources