Self Hosted WhatsApp API: Run Your Own Server Gateway for Free

OpenWA gives developers a free, self-hosted WhatsApp API in one Docker command

OpenWA is a free, MIT-licensed WhatsApp API gateway that developers can run on their own infrastructure. It supports sending and receiving messages, multiple WhatsApp sessions, webhooks, a built-in dashboard, and integrations with automation tools such as n8n and Home Assistant. The project is Docker-native, so a local setup can be launched with a single command and managed through its web interface. It's designed to give teams more control over their messaging stack without subscription fees or an outside API provider.

• Self-hosting gives developers direct control over message infrastructure, data storage, and automation logic.
• Teams should still review WhatsApp's terms and policies before using any unofficial integration in production.

RIVR is testing delivery robots where real-world conditions get messy

Amazon-backed robotics company RIVR is testing its autonomous delivery robots in extreme weather and difficult terrain. The work is intended to measure how the robots handle rain, snow, uneven ground, and other unpredictable conditions that make last-mile delivery difficult outside controlled pilot routes. RIVR's robots combine wheeled mobility with legged capabilities, aiming to navigate the final stretch from a delivery vehicle to a customer's door.

• Reliability in bad weather and on uneven terrain is one of the biggest hurdles before autonomous delivery can scale.
• Passing real-world stress tests matters more than polished sidewalk demos for proving commercial readiness.

A developer ran a 28.9M-parameter language model on an $8 microcontroller

Developer slvDev ran a 28.9 million-parameter TinyStories language model entirely offline on an ESP32-S3 microcontroller, a chip class that costs roughly $8. The model produces text at around 9.5 tokens per second despite the chip having only 512KB of SRAM.

The project uses 4-bit quantization and per-layer embeddings, keeping about 25 million parameters in 16MB of flash while the compute core runs in SRAM and working memory uses PSRAM. The full quantized model is about 14.9MB, and the project is open source.

• Flash storage makes it possible to fit a model far larger than the microcontroller's working memory.
• Small local models could enable cheap, private AI controls for sensors, appliances, and embedded devices without a cloud connection.

Precision laser cleaning can remove paint and markings without touching the surface

A Chinese laser-equipment maker has demonstrated precision cleaning systems that use focused laser energy to remove paint, markings, rust, oil, coatings, and other contaminants while preserving the underlying substrate. Unlike abrasive blasting or chemical stripping, laser cleaning is non-contact and highly controllable, allowing operators to target a specific layer or area.

These systems are used across manufacturing, maintenance, restoration, automotive work, and surface preparation, with pulsed-laser machines offering the greatest control for delicate materials.

• Precise, non-contact cleaning can reduce material wear and eliminate the chemical waste associated with stripping methods.
• Results depend on laser power, wavelength, pulse settings, and the material being cleaned, so expert calibration remains essential.

Self-hosting OpenWA with Docker Compose

While the official WhatsApp Business Cloud API requires Meta Business verification, credit cards, and per-conversation fees, open-source gateways like OpenWA let you spin up a self-hosted REST API on any cheap VPS (such as a $4/month Hetzner or DigitalOcean droplet). Here is a production-ready docker-compose.yml configuration:

version: '3.8'

services:
  openwa:
    image: openwa/wa-automate:latest
    container_name: openwa-gateway
    restart: unless-stopped
    ports:
      - "127.0.0.1:8080:8080"
    environment:
      - WA_API_KEY=your_super_secret_api_token_here
      - WA_WEBHOOK_URL=http://n8n:5678/webhook/whatsapp-incoming
      - WA_POPUP=false
      - WA_HEADLESS=true
      - WA_AUTO_REFRESH=true
    volumes:
      - ./sessions:/app/session
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
      interval: 30s
      timeout: 10s
      retries: 3

Authentication and session persistence

Getting initial pairing and maintaining session stability requires specific precautions:

• Initial QR Pairing: Launch the container with logs visible (docker compose logs -f). Scan the generated terminal ASCII QR code using WhatsApp on your mobile device (Linked Devices → Link a Device).
• Persistent Volume Mounts: Always mount /app/session to a local host directory. WhatsApp sessions store cryptographic keys locally; if your container rebuilds without this volume, you will be logged out and forced to re-scan.
• Security Hardening: Never expose port 8080 directly to the public internet without an API key or reverse proxy. Bind the port to 127.0.0.1 and proxy it behind Caddy or Nginx with SSL and strict IP allowlisting for your webhook endpoints (like n8n).

Rate limits and ban prevention strategies

Meta actively monitors automated patterns on client-driven accounts. To prevent account bans or number flagging, implement these operational safeguards: