ApisNode is a production-deployed smart beehive monitor — 30 real-time metrics per hive (environmental, acoustic, vibration, and gas/VOC fingerprint), AI-powered daily colony health reports, and over-the-air firmware updates for precision apiculture.
🌐 Live at apisnode.com — includes a live telemetry dashboard pulling data from production hives and an AI beekeeping assistant powered by Amazon Bedrock.
🐝 Open Apiary Project — our proposed 501(c)(3) non-profit initiative deploying free ApisNode beehive monitors to beekeepers nationwide to combat Colony Collapse Disorder.
ApisNode continuously monitors beehive health using a custom ESP32-C6 sensor node installed inside the hive. Every wake cycle it captures 30 real-time metrics — temperature, humidity, barometric pressure, hive tilt, a 10-band acoustic spectrum, a 6-band vibration spectrum, and a 10-step gas (VOC) fingerprint — transmits them via encrypted LoRa radio to an always-on gateway, and relays them to the cloud for AI-powered daily colony health analysis.
No WiFi needed at the hive. No manual data collection. No interrupting the bees.
┌─────────────────────────────────────────────────────────────────┐
│ APIARY (Edge Hardware) │
│ │
│ ┌─────────────────┐ LoRa 916 MHz ┌──────────────────────┐ │
│ │ ApisNode │ AES-128-GCM ──► │ ApisMind │ │
│ │ (In-Hive) │ 70B / 60s │ (Gateway) │ │
│ │ ESP32-C6 │ ◄── CONFIG ────── │ ESP32-C6 │ │
│ │ Battery-Powered│ │ Wall-Powered │ │
│ └─────────────────┘ └──────────┬───────────┘ │
└────────────────────────────────────────────────────┼────────────┘
│ WiFi / MQTT / TLS
┌──────────▼───────────┐
│ AWS Cloud │
│ IoT Core → DynamoDB │
│ Lambda → Bedrock │
│ (Claude 4.5 Haiku) │
└──────────┬───────────┘
┌────────────────────────────┼────────────────────────────┐
│ │ │
┌────────▼────────┐ ┌──────────▼──────────┐ ┌──────────▼──────────┐
│ Mobile App │ │ Wall Display │ │ Fleet Dashboard │
│ Flutter │ │ display.apisnode.com │ │ fleet.apisnode.com │
│ iOS / Android │ │ Live Spectrograms │ │ OTA Management │
└─────────────────┘ └─────────────────────┘ └─────────────────────┘
Two complementary ESP32-C6 devices work together at each apiary:
Battery-powered, installed between frames. Spends nearly all time in deep sleep, waking every 60 seconds for a Sense → DSP → Transmit → Sleep cycle.
| Sensor | Measurements |
|---|---|
| Bosch BME688 | Temperature, humidity, barometric pressure, 10-step gas (VOC) heater-scan fingerprint |
| ADXL362 Accelerometer | 6-band vibration spectral analysis (10–1000 Hz), hive tilt |
| Knowles PDM Microphone | 10-band audio spectral analysis (100–3600 Hz), colony sound level |
Wall-powered, installed outside the apiary. Runs a continuous LoRa receiver — no polling or timing synchronization needed. Bridges the apiary radio network to AWS IoT Core over WiFi/MQTT/TLS and provides an outdoor environmental baseline for AI weather correlation.
Also handles BLE provisioning (WiFi credentials, IoT certificates, LoRa config) and store-and-forward resilience via SD card queue when connectivity drops.
Every wake cycle, ApisNode runs on-chip FFT-based feature extraction with 5× burst averaging for stability:
| Band | Range | What It Detects |
|---|---|---|
| 0 | 100–150 Hz | Baseline colony hum, pre-swarm energy build |
| 1 | 150–200 Hz | Worker piping, pre-swarm frequency shift |
| 2 | 200–260 Hz | Swarming indicator, waggle dance overlap |
| 3 | 260–350 Hz | Colony fundamental tone, queen quacking |
| 4 | 350–420 Hz | Queen quacking → tooting transition |
| 5 | 420–500 Hz | Queen tooting (piping), active swarming |
| 6 | 500–600 Hz | Full-spectrum swarming excitation |
| 7 | 600–800 Hz | Harmonics, broadband distress |
| 8 | 800–1500 Hz | Hissing, defensive behavior |
| 9 | 1500–3600 Hz | Broadband hissing (upper range) |
| Band | Range | What It Detects |
|---|---|---|
| 0 | 10–50 Hz | External disturbance, mechanical impact |
| 1 | 50–100 Hz | Colony metabolic baseline |
| 2 | 100–200 Hz | Transitional activity, lower harmonics |
| 3 | 200–300 Hz | Waggle dance (primary foraging diagnostic) |
| 4 | 300–500 Hz | Queen presence indicator |
| 5 | 500–1000 Hz | Broadband stress, disease markers |
Per-Band AGC: Each frequency band independently tracks its own baseline via exponential moving average, ensuring every band maintains consistent visibility regardless of absolute energy level.
The BME688's gas sensor is stepped through 10 heater setpoints (100, 150, 200, 250, 280, 300, 320, 350, 375, 400 °C) in forced mode each cycle, building a volatile-organic-compound fingerprint far richer than a single gas reading. Temperature and humidity are sampled first with the heater off, so the ambient readings stay clean.
Shifts in this 10-point profile — tracked against the colony's own baseline — flag changes in hive air chemistry: nectar curing, propolis and brood volatiles, or contamination worth a closer look at the next inspection. Each step is log₂-encoded against a shared reference for compact LoRa transmission.
A daily Lambda (EventBridge-triggered) fuses multiple data sources and calls Amazon Bedrock (Claude 4.5 Haiku) to generate per-hive health reports:
┌──────────────────────────────────────────────────────────┐
│ LLM Prompt Context │
├───────────────────┬─────────────────────────────────────-┤
│ Bee Science KB │ Curated peer-reviewed research by │
│ │ sensor modality (acoustics, vibration,│
│ │ air quality, temperature, seasonal) │
├───────────────────┼──────────────────────────────────────┤
│ 24h Telemetry │ In-hive sensors + outdoor ApisMind │
│ │ weather baseline (hyper-local) │
├───────────────────┼──────────────────────────────────────┤
│ Inspection Notes │ 7-day voice-transcribed field │
│ │ observations + AI-extracted labels │
├───────────────────┼──────────────────────────────────────┤
│ Prior Analysis │ Previous health scores, detected │
│ │ issues, and recommended actions │
└───────────────────┴──────────────────────────────────────┘
Output per hive: health score (0–100), status (healthy/warning/critical), detected issues, and actionable recommendations — surfaced in the mobile app.
All in-apiary communication uses LoRa sub-GHz radio with hardware encryption:
| Parameter | Value |
|---|---|
| Frequency | 916 MHz (US915 quiet channel) |
| Spreading Factor | SF10 |
| Bandwidth | 125 kHz |
| TX Power | 14 dBm |
| Encryption | AES-128-GCM (4-byte tag, fixed IV) |
| Packet Size | 70 bytes per reading |
| TX Interval | 60 seconds (configurable) |
New ApisNodes auto-pair with ApisMind on first boot. The gateway enforces a −20 dBm RSSI threshold — the node must be within ~1–2 feet of the gateway to exchange keys. No manual configuration required.
Both devices support remote over-the-air firmware updates using binary delta patches:
| Path | Typical Patch Size | Transfer Time |
|---|---|---|
| ApisMind (WiFi) | 20–90 KB | ~12 seconds |
| ApisNode (LoRa relay) | 8–30 KB | ~90 seconds |
Deployed via AWS CDK:
- IoT Core — MQTT broker receiving encrypted telemetry from ApisMind gateways
- DynamoDB — Telemetry store with TTL lifecycle and adaptive time-bucketed downsampling
- Lambda — REST API (API Gateway), daily AI analysis, inspection labeling, fleet management
- Amazon Bedrock — Claude 4.5 Haiku for daily colony health reports and interactive chat
- Cognito — JWT authentication with server-side device ownership enforcement
- S3 — Firmware binaries, OTA delta patches, inspection audio/transcripts
- Amplify — Web dashboard hosting with automatic deploys
BLE device provisioning, animated 30-metric dashboard, real-time spectrograms, AI health reports with interactive chat, voice-to-label hive inspections, remote device management (reboot, recalibration, OTA triggers).
🖥️ Wall Display — display.apisnode.com
Always-on kiosk dashboard for wall-mounted screens. Live mini-graphs, 2D spectrograms with date/time axes, battery history, 1H–30D time range selector. Timestamp-based X positioning makes offline gaps visually apparent.
📊 Fleet Dashboard — fleet.apisnode.com
Admin fleet monitoring — firmware version tracking, OTA deployment management, device health overview.
🌐 ApisNode Website — apisnode.com
Public product website featuring a live telemetry dashboard from production hives, a live AI beekeeping chat grounded in real sensor data, and daily published AI colony health reports from monitored apiaries across New Mexico, Florida, and Montana.
Open Apiary Project — openapiaryproject.org
The Open Apiary Project is our proposed 501(c)(3) non-profit initiative dedicated to combating Colony Collapse Disorder.
The mission: Deploy free ApisNode beehive monitor kits to beekeepers nationwide — building the world's most comprehensive anonymized apiary health database to support CCD research, pollination forecasting, and regional disease outbreak detection.
US beekeepers lose an average of 40–50% of managed colonies per year. The pollinators responsible for one-third of the US food supply face an existential threat. Continuous, non-invasive monitoring is a critical tool for early detection and intervention.
Learn more and get involved → openapiaryproject.org
| Layer | Technology |
|---|---|
| Firmware | ESP-IDF v5.5.1, C/C++, FreeRTOS |
| Radio | Seeed Wio-E5 (STM32WLE5), LoRa AT commands |
| Sensors | Bosch BME688 (SPI), ADXL362 (SPI), Knowles PDM (I2S) |
| DSP | On-chip FFT, per-band AGC via EMA, 10-step BME688 gas heater scan |
| Mobile | Flutter 3.x (Dart), iOS & Android |
| Cloud | AWS CDK v2, Lambda, DynamoDB, IoT Core, Bedrock, Cognito, S3, Amplify |
| AI | Amazon Bedrock — Claude 4.5 Haiku |
| OTA | detools delta patches, heatshrink compression, SHA256, ESP-IDF rollback |
| Web | Vanilla HTML/CSS/JS, AWS Amplify |
Proprietary — © Acid Canyon LLC. All rights reserved.
The source code for this project is closed-source. This repository serves as a public technical overview.
For inquiries: apisnode.com · openapiaryproject.org