OpenClaw + Supermemory: Give Your AI Long-Term Memory

By default, OpenClaw stores memory as plain Markdown files on disk. It works, but it's limited — search is basic, there's no semantic understanding, and scaling is manual.

Supermemory changes this. It gives OpenClaw a cloud-powered memory system that automatically captures, indexes, and recalls information with semantic search.

What Supermemory Does

• Auto-capture: Every conversation is automatically stored
• Smart recall: Before each AI turn, relevant memories are injected as context
• Semantic search: Find memories by meaning, not just keyword matching
• Cross-session: Memory persists across conversations and restarts
• Visualization: View your knowledge graph to see how memories connect

How It Works

1. You chat with your OpenClaw bot on Telegram
2. Supermemory stores the conversation after each message
3. On your next message, Supermemory retrieves relevant past context
4. OpenClaw receives this context alongside your new message
5. The AI responds with full awareness of your history

The result: your assistant knows what you discussed last week, remembers your preferences, and builds on previous conversations.

Setup: Self-Hosted

Prerequisites

• Supermemory Pro plan ($29/mo) — sign up at console.supermemory.ai
• Your Supermemory API key (starts with sm_)

Install the Plugin

openclaw plugins install @supermemory/openclaw-supermemory

Configure

Add your API key to the environment:

export SUPERMEMORY_OPENCLAW_API_KEY=sm_your_key_here

Set the container tag for user isolation:

openclaw config set plugins.entries.openclaw-supermemory.config.containerTag "my_user"

Restart OpenClaw:

openclaw restart

Verify

Send a few messages to your bot, then check the Supermemory dashboard at console.supermemory.ai to see captured memories.

Setup: ClawTank

On ClawTank, Supermemory is pre-configured:

1. The plugin is automatically installed during container provisioning
2. Memory isolation is handled per-user (using containerTag)
3. The Memory page in your dashboard shows an interactive knowledge graph

No manual configuration needed — just start chatting and your memories build automatically.

The Memory Graph

ClawTank includes a visual memory graph at /dashboard/memory. This interactive visualization shows:

• All stored memories as nodes
• Connections between related memories
• Timeline of when memories were created
• Search across all your stored knowledge

It's useful for understanding what your assistant knows about you and verifying that important context has been captured.

Tips for Better Memory

Be Explicit About Important Information

Instead of casually mentioning something, tell your assistant directly:

"Remember: my client John prefers email over Slack, and his timezone is PST"

Use the /remember Command

OpenClaw has a built-in /remember command that explicitly stores information:

"/remember My budget for the Q2 marketing campaign is $15,000"

Review and Clean Up

Periodically check your memory graph and remove outdated information. Old context can confuse the AI if circumstances have changed.

Memory Survives Rebuilds

An important detail: on ClawTank, memory is tied to your user account, not your container. If you rebuild your OpenClaw instance, all memories are preserved.

This is because Supermemory uses a containerTag based on your user ID — the cloud-stored memories don't depend on the Docker container.

Cost

Supermemory Pro is $29/month. For most users, this is worth it — persistent memory transforms OpenClaw from a stateless chatbot into a true personal assistant that knows you.

Get Started

Deploy OpenClaw with Supermemory pre-configured on ClawTank. Your assistant starts building memory from the first conversation.