Details

Turn documents, sheets, specs and notes into something you can query. Ask "when did we change this and why?" and get an answer with sources — not guesses.

Content is ingested, dynamically tagged (topics, people, decisions etc) and stored as a graph. You can explore, ask, interact and trace every answer back to source.

Details

AI-assisted writing environment for teams that need to create content fast without losing their own tone. Built to turn ideas into publishable drafts instead of just generating generic AI text.

Inline editing actions (rewrite, extend, translate, summarize, check grammar) built directly into the editor. Custom brand voice and private knowledge base so the assistant writes in your style, using your material. I designed the product, visual identity, and core UX, and implemented the working interface and LLM integration in Next.js.

Details

AI grocery planner for Rema1000. Helps you pick what to cook, turns it into a structured shopping list, and links every ingredient to real products with live prices and one-tap add-to-cart.

It keeps separate lists per meal or use case, can turn a user photo into a shopping list, shows exact quantities, and maps each ingredient to the correct in-store SKU and price. It can also answer practical questions about Rema1000 and more.

Details

Automated data extraction pipeline that pulls live data from external sources, cleans it, and stores it in a structured format for use in internal tools.

This scraper collects data from multiple sources, deduplicates it, enriches it with LLM-based cleanup, and exposes it as clean, queryable data through a dashboard. It is one of several scraping solutions I’ve delivered for different needs.

Details

Location insight tool I built for selling my house. You enter an address and instantly see what’s around it — schools, grocery stores, playgrounds, green areas, transport — visualized on a map.

The map pulls open public data instead of paid map APIs, plots relevant points around the property, and lets you filter what matters to a buyer. I built it so potential buyers could understand the value of the area immediately.

Details

Personal AI assistant that can chat, fetch live information, control my home, take notes, and generate media.

It can call tools for weather and Wikipedia, operate my Home Assistant setup, create new Google Docs notes, generate images locally and audio, accept voice input, and reply with spoken answers. Built as my own local and unified assistant instead of using separate apps.

Details

WordPress plugin that takes uploaded photos (custom post type) and uses an LLM to generate platform-ready posts — captions, tone, hashtags, and tags for each social platform.

It schedules and auto-publishes that generated content to multiple SoMe platforms on an interval. All setup — accounts, timing, and post styling — is managed directly in WordPress, no third-party tool needed.

Details

A production-grade design system and UI kit: marketing pages, dashboards, components, tokens, and documentation in one Figma workspace.

The UI studio is a full-scale design system built for real products — not just pretty shots. It includes responsive page sections, dashboard UI, component states, spacing, typography, tokens, and usage guidelines so teams can move from idea to interface fast.

Details

Figma plugin that utilizes The UI studio designsystem to assemble interfaces automatically — no blank canvas, no manual wiring.

The Figma plugin can drop in full page sections, apply the correct spacing, typography, color theme, and adjust copy placeholders and images.

Details

The largest library of avatars in the world. Customizable and free for commercial use.

Avaaatars.com is a large, free avatar library for product shots, dashboards, decks, and marketing. You can pick a style, tweak look and mood, and export assets for commercial use with no licensing friction.

Details

Full-scale replica of Elena of Avalor’s scepter, built for my daughter. It can record and cast ‘spells’ that trigger real actions at home, with reactive light, sound, motion gestures, and an LLM that answers in the character’s voice.

The scepter has LEDs in the crystal, onboard speaker for character voice lines, motion sensing for gesture-based spells, and Home Assistant integration so gestures can control lights and other devices in the house. It also has a voice-triggered LLM that responds in the character’s voice.

Details

A universal remote built on ESP32 that can learn IR signals from existing remotes and replay them to control TVs, lights, receivers, and more.

The ESP32 stores captured IR codes and serves a local web interface where I can manage signals, create actions, and set up automations (like multi-step power/input macros).

Details

Standalone voice assistant device with local speech recognition, text-to-speech, and a small LLM that can control my home and answer questions.

The device listens, transcribes, routes intent to a local model, can trigger Home Assistant automations, call APIs, create notes, and then talks back using onboard speech. Built as a fully voice-first interface, not a screen app.