What cloud subscription gets you
The open source core is fully functional—you can run everything yourself forever. But if you'd rather focus on growing than managing infrastructure, here's what our cloud adds.
Zero infrastructure management
No server to maintain, no database to back up, no security patches to apply. We handle uptime, scaling, and updates so you can focus on your grow.
ML-powered irrigation
Our machine learning models analyze your sensor patterns, environmental conditions, and historical data to predict optimal irrigation timing—beyond simple threshold triggers.
Future LLM integrations
Coming soon: Ask questions about your grow in natural language, get AI-generated insights, and receive intelligent recommendations based on your specific setup.
Push notifications & alerts
Instant alerts when something needs attention—reservoir low, sensor offline, unusual readings. Managed notification infrastructure you don't have to set up.
Self-hosters can access ML features too
Running your own infrastructure but want our ML predictions? No problem. Self-hosted instances can connect to our ML API for intelligent irrigation recommendations while keeping all your raw data local. Pay only for what you use.
Your subscription funds ongoing development of both the cloud service and the open source core. When you pay us, everyone benefits.
Everything you need is open source
We're not open-washing with a 'community edition' that's missing critical features. The full system—hardware, firmware, backend, dashboard—is open source and fully functional.
Our cloud runs on this exact code
This isn't a separate "community edition"—our hosted cloud service is built directly on top of these same open source repositories. When we improve the cloud, the open source core improves. When you contribute a fix, cloud users benefit too.
ESP32 Firmware
AGPL-3.0The complete firmware that runs on your controller. Sensor reading, valve control, MQTT communication, WiFi provisioning, and offline failsafe logic.
- Full source code on GitHub
- Build and flash yourself
- Modify for your hardware
- Works with any MQTT broker
- No cloud credentials baked in
Decision Service (Backend)
AGPL-3.0Python/FastAPI backend that ingests sensor data, makes irrigation decisions, and serves the dashboard API. Everything you need to run Phyt on your own infrastructure.
- Full REST API
- MQTT integration
- TimescaleDB for time-series data
- User authentication
- Docker Compose deployment
Dashboard (Frontend)
AGPL-3.0React + Vite web application for monitoring and controlling your grow. Real-time sensor data, zone configuration, historical charts, and device management.
- Modern React with TypeScript
- Mobile-responsive design
- Real-time updates via WebSocket
- Self-hostable static files
- No tracking or analytics
Hardware Designs
CERN-OHL-S-2.0 (planned)Schematics, PCB layouts, and enclosure designs for the controller and valve modules. Currently in beta refinement—will be released once stable.
- KiCad schematics
- PCB Gerber files
- BOM (Bill of Materials)
- 3D printable enclosures
- Assembly instructions
This isn't marketing fluff
Open source isn't just a checkbox for us. It's a fundamental commitment to growers who've been burned by vendor lock-in before.
Your grow never depends on us
If Phyt the company disappears tomorrow, your system keeps running. You have the source code. You can maintain it yourself, fork it, or find someone else to help.
Complete data ownership
Self-hosters keep all data on their own hardware. No data ever leaves your network. Perfect for commercial grows or anyone who values privacy.
Customize everything
Need to support a custom sensor? Add a feature we haven't built? Fork the repo and make it yours. No waiting for us to prioritize your request.
No hostage situations
Cloud users can export all their data anytime and switch to self-hosting. We don't hold your historical data hostage to keep you subscribed.
"If you can't run it yourself, you don't own it—you're just renting it."
Our philosophy on grow automation
Self-host vs. Cloud
Both options use the same open source core. The difference is who manages the infrastructure and what convenience features are included.
| Feature | Self-HostFree forever | Cloud$12-99/mo |
|---|---|---|
| Full firmware source code | ||
| Full backend source code | ||
| Full dashboard source code | ||
| Threshold-based irrigation | ||
| Unlimited devices | Tier-dependent | |
| Unlimited data retention | Tier-dependent | |
| 100% local data | ||
| Works offline | ||
| Data export anytime | N/A (you own it) | |
| ML-powered irrigation | Via API ($) | |
| LLM insights (coming) | Via API ($) | |
| Zero infrastructure management | ||
| Managed updates | ||
| Push notifications | DIY | |
| Priority support |
Cloud users can switch to self-hosting anytime. Export your data, point your devices at your own MQTT broker, and you're free. No penalty, no data loss.
Run it on a Raspberry Pi
Everything you need to run Phyt locally fits on a $50 Raspberry Pi. Here's what the self-hosted stack looks like.
# Clone the repo
git clone https://github.com/phyt-ai/phyt.git
cd phyt
# Start the stack (PostgreSQL + Mosquitto + Backend + Dashboard)
docker-compose up -d
# Flash firmware to your ESP32
cd firmware
pio run -t upload
# Provision your device
# 1. Connect to "Phyt-Setup-XXXX" WiFi
# 2. Enter your home WiFi credentials
# 3. Point to your local MQTT broker (e.g., raspberrypi.local:1883)
# Access your dashboard
open http://raspberrypi.local:5173~500MB
Total disk usage
~200MB
RAM usage (idle)
5 min
Setup time
Why AGPL?
We chose the GNU Affero General Public License (AGPL-3.0) for our software because it ensures that improvements come back to the community.
What AGPL means for you:
- Personal/commercial use: Run it for your own grow, commercial or hobby. No restrictions.
- Modify freely: Change the code however you want for your own use.
- Distribute modifications: If you distribute modified versions or run them as a service for others, you must share your changes under AGPL.
For hardware, we plan to use the CERN Open Hardware License, which follows similar principles for physical designs.