Platform vs Community Edition
Understand the differences between Engram Cloud and self-hosted community edition
Platform vs Community Edition
Engram ships in two forms: a self-hosted Community Edition and the hosted Platform (Cloud). They serve different needs and can work together.
Community Edition
A single Docker container that bundles everything needed for fully local AI memory:
docker pull engrammemory/engram-stack
docker run -p 8585:8585 -p 6333:6333 -p 11435:11435 -v engram-data:/data engrammemory/engram-stack
What's inside:
- Vector database
- Local embedding engine
- Recall engine with MCP server
- Zero external API calls
MCP Tools (6):
| Tool | Description |
|---|---|
memory_store | Store a new memory |
memory_search | Semantic search across memories |
memory_recall | Retrieve specific memories by ID or context |
memory_forget | Delete memories |
memory_consolidate | Merge and compress related memories |
memory_connect | Link related memories together |
No API key. No billing. No data leaves your machine.
Platform (Cloud)
The hosted intelligence API at https://api.engrammemory.ai/v1.
What you get beyond Community:
- Proprietary compression (6-8x vector size reduction)
- Auto-classification pipeline
- Multi-tier intelligent recall (sub-millisecond)
- Python and JavaScript SDKs
- Webhooks and event triggers
- Edge device fleet management
- Compliance and audit logging
- Usage analytics dashboard
Feature Comparison
| Feature | Community | Free | Builder | Scale | Enterprise |
|---|---|---|---|---|---|
| Memory store/search/recall | Yes | Yes | Yes | Yes | Yes |
| MCP server (6 tools) | Yes | — | — | — | — |
/v1/intelligence pipeline | — | Yes | Yes | Yes | Yes |
| Proprietary compression | — | Yes | Yes | Yes | Yes |
| Auto-classification | — | Yes | Yes | Yes | Yes |
| Intelligent recall | — | Yes | Yes | Yes | Yes |
| Python & JS SDKs | — | Yes | Yes | Yes | Yes |
| Webhooks | — | — | Yes | Yes | Yes |
| Triggers | — | — | Yes | Yes | Yes |
| Edge device management | — | — | — | Yes | Yes |
| Compliance & audit logs | — | — | — | — | Yes |
| SSO / SAML | — | — | — | — | Yes |
| Dedicated support | — | — | — | Priority | Dedicated |
| Token limit / month | Unlimited | 100K | 2M | 25M | Unlimited |
| Query limit / month | Unlimited | 1K | 20K | 250K | Unlimited |
When to Use Which
Choose Community when:
- Privacy is non-negotiable and no data can leave your network
- You need offline or air-gapped operation
- You are building MCP-based integrations (Claude Desktop, etc.)
- You want zero ongoing costs
Choose Platform when:
- You need production-grade compression and deduplication
- You are building with the SDKs or REST API
- You need webhooks, triggers, or fleet management
- You want managed infrastructure with usage analytics
- You need compliance features or audit trails
Using Them Together
Community Edition can overflow to Platform when local storage or compute is insufficient. Add your Platform API key to the Community config:
# community-config.yaml
overflow:
enabled: true
api_key: "pr_live_xxxxx"
threshold_mb: 500 # Overflow to Platform when local storage exceeds 500MB
Memories stay local until the threshold is hit. Overflow memories are processed through the full Platform intelligence pipeline (compression, dedup, classification) before storage.
This gives you the best of both: local-first privacy with Platform-scale capacity as a fallback.
Next Steps
- Quickstart — Get started with Platform in 5 minutes
- Self-Hosted Guide — Deploy and configure Community Edition
- API Reference — Full Platform API documentation