w3cj writes about jev-chat, a tool-calling chat bot that routes user requests to real tools using Jev, a non-generative classifier from TypeSafe, with no LLM writing any output. Every value on screen was either typed by the user or returned by a tool, so the assistant cannot invent a fact. The system supports weather, unit conversion, Wikipedia lookups, recipes, web search, Todoist, and Home Assistant via MCP servers, with an inspector pane exposing every decision and probability for each turn.
- Jev answers only two question types — Choice and Noul — and never produces text; all reply wording is templated in code
- Pre-processing handles spell-check (cspell + compromise) and resolves short follow-ups like "what about Boston?" by swapping in the new value
- Multi-step tools (Wikipedia, web search) chain multiple Jev requests: pick topic, then article, then the exact line that answers
- Confidence-gated routing shows two buttons when the top tools are close rather than guessing
- The repo is a proof of concept; the author will not accept PRs for new features
- English only; no compound requests or multi-step reasoning supported
- Jev answers only two question types — Choice and Noul — and never produces text; all reply wording is templated in code
- Pre-processing handles spell-check (cspell + compromise) and resolves short follow-ups like "what about Boston?" by swapping in the new value
- Multi-step tools (Wikipedia, web search) chain multiple Jev requests: pick topic, then article, then the exact line that answers
- Confidence-gated routing shows two buttons when the top tools are close rather than guessing
- The repo is a proof of concept; the author will not accept PRs for new features
- English only; no compound requests or multi-step reasoning supported
Rich Edmonds writes about transitioning from subscription-based smart home services like Ring to a self-hosted local NVR system using Frigate. By moving video storage and AI detection onto his own hardware instead of relying on third-party clouds, he eliminated annual fees while gaining greater control over privacy and data retention.
- The author saved $85 per year by ditching Reolink ($35) and Ring ($50) subscriptions.
- A self-hosted setup using an old PC with Frigate can provide much longer video retention than cloud services.
- Building a custom NVR is not entirely free; the author's dedicated server consumes about 160 watts of power.
- Home Assistant integration allows for advanced features like Alarmo and LLM vision analysis via local models.
- The author saved $85 per year by ditching Reolink ($35) and Ring ($50) subscriptions.
- A self-hosted setup using an old PC with Frigate can provide much longer video retention than cloud services.
- Building a custom NVR is not entirely free; the author's dedicated server consumes about 160 watts of power.
- Home Assistant integration allows for advanced features like Alarmo and LLM vision analysis via local models.
Tim Brookes writes about seven projects you can build with the Cheap Yellow Display (CYD), an affordable ESP32 board with an integrated LCD touch screen that lets you skip soldering and breadboards entirely. The projects span a weather station, a Wi-Fi and Bluetooth penetration testing suite (ESP32 Marauder), a Bluetooth MP3 player, a retro mini TV for videos and games, a Gran Turismo 7 racing dashboard, a Home Assistant music remote, and an ESPHome smart home controller.
- The weather station pulls from both Open-Meteo and OpenWeatherMap and supports a DHT22 sensor for local temperature readings
- The ESP32 Marauder CYD port covers hardware from 1.9-inch non-touch models up to 3.5-inch touch displays
- The retro mini TV is a 3D-printed Simpsons replica combining the Anemoia-ESP32 NES emulator and MiniLegoTV video player
- The GT7 dashboard works because Gran Turismo 7 publishes telemetry over UDP on the local network, a feature shared by other racing simulators
- The CYD Jukebox pairs with Music Assistant to unify local files and streaming services like Apple Music behind one interface
- The weather station pulls from both Open-Meteo and OpenWeatherMap and supports a DHT22 sensor for local temperature readings
- The ESP32 Marauder CYD port covers hardware from 1.9-inch non-touch models up to 3.5-inch touch displays
- The retro mini TV is a 3D-printed Simpsons replica combining the Anemoia-ESP32 NES emulator and MiniLegoTV video player
- The GT7 dashboard works because Gran Turismo 7 publishes telemetry over UDP on the local network, a feature shared by other racing simulators
- The CYD Jukebox pairs with Music Assistant to unify local files and streaming services like Apple Music behind one interface
Tanveer Singh:
- microSD desk photo frame
- PC system stats monitor
- Home Assistant dashboard
- Wi-Fi and Bluetooth security tool
- ASCII aquarium
- microSD desk photo frame
- PC system stats monitor
- Home Assistant dashboard
- Wi-Fi and Bluetooth security tool
- ASCII aquarium
Building an affordable Home Assistant dashboard does not require expensive hardware. This guide explores several low-cost methods to create functional control panels and information displays using existing or cheap equipment.
Ways to build a budget dashboard include:
* Repurposing old tablets, phones, or even Kindles from your junk drawer using kiosk browsers or the Home Assistant companion app.
* Purchasing second-hand hardware like older tablets or digital photo frames through online marketplaces.
* Using small ESP32 touchscreen modules for low-power desktop dashboards.
* Implementing E-Ink displays for minimal, always-on information such as weather and calendars.
* Jailbreaking used smart displays to bypass manufacturer limitations and run custom software.
Ways to build a budget dashboard include:
* Repurposing old tablets, phones, or even Kindles from your junk drawer using kiosk browsers or the Home Assistant companion app.
* Purchasing second-hand hardware like older tablets or digital photo frames through online marketplaces.
* Using small ESP32 touchscreen modules for low-power desktop dashboards.
* Implementing E-Ink displays for minimal, always-on information such as weather and calendars.
* Jailbreaking used smart displays to bypass manufacturer limitations and run custom software.
An exploration of an experiment involving connecting a local Large Language Model to Home Assistant to control a smart light bulb. By assigning the AI a specific persona through custom system prompts, the author attempted to make the lighting respond emotionally to environmental data. While successful in creating reactive lighting, the experience ultimately became unsettling as the model made autonomous decisions without direct input.
- Connecting local LLMs via LM Studio and Home Assistant
- Using system prompts to define device personalities
- Automating smart bulb color and brightness through AI reasoning
- The psychological impact of unsupervised AI autonomy in a smart home environment
- Connecting local LLMs via LM Studio and Home Assistant
- Using system prompts to define device personalities
- Automating smart bulb color and brightness through AI reasoning
- The psychological impact of unsupervised AI autonomy in a smart home environment
An exploration of using Claude Code to develop a Google Nest Hub-like device with an ESP32-P4, highlighting the challenges and limitations of AI-assisted development for niche hardware.
Pixlpal is a hackable, ESP32-S3-based desktop device with an 11.25-inch LED matrix, high-fidelity audio, and Home Assistant integration, designed to be a smart AIoT desktop companion.
Discussion thread about using a Seeed Studio reTerminal E1002 color ePaper display as a Home Assistant dashboard.
This article details how to run a 120B parameter LLM locally with 24GB of VRAM and 64GB of system RAM, using a setup with Proxmox LXCs, Whisper for voice transcription, and integration with Home Assistant for smart home automation.
