Get persistent AI memory running in 15 minutes.
- Zo Computer with root access
- ~500MB free disk space
curl -sSL https://raw.githubusercontent.com/theforkproject-dev/zo-local-memory/main/install.sh | bashThis installs:
- Ollama (embedding service)
- Turso/sqld (vector database)
- Python memory client libraries
Critical: Your AI needs a special persona configuration to use memory.
- Go to Settings > Your AI > Personas
- Click "Create Persona"
- Name: "Memory-Enabled Zo"
- Copy the complete persona prompt from PERSONA.md
- Save and "Set as Active"
With your Memory-Enabled Zo persona active, start chatting. The AI should automatically initialize memory at the start.
You: "I prefer concise technical explanations without verbose context."
The AI will autonomously store this as a memory.
cd /home/workspace/.zo
python3 local_memory_client.py search "technical explanations" 3 vectorYou should see your preference stored.
Start a new conversation and ask:
You: "What do you remember about my communication preferences?"
The AI should recall and reference your stored preference.
✅ Persistent memory - AI remembers across sessions
✅ Semantic search - Find memories by meaning, not exact words
✅ Full data sovereignty - Everything runs locally
✅ Zero API costs - No external dependencies
✅ Session continuity - Conversations build on each other
- Full deployment guide: DEPLOY.md
- Persona configuration details: PERSONA.md
- Architecture deep-dive: ARCHITECTURE.md
- Tools hub: https://tools-hub-fork.zocomputer.io/local-memory
Solution: Verify persona is active (not just created). Look for "Set as Active" button in persona settings.
Solution: Ensure you're running on a Zo Computer. The installer checks for /etc/zo/supervisord-user.conf.
Solution:
supervisorctl -c /etc/zo/supervisord-user.conf status ollama
supervisorctl -c /etc/zo/supervisord-user.conf status sqld
tail /dev/shm/ollama_err.log
tail /dev/shm/sqld_err.logHaving trouble? Open an issue on GitHub