Tools
Small apps I built for my own use — they run on the Mac, keep your data local, and solve one problem each.
These tools were built for personal use and live in private GitHub repositories. You're welcome to request access — just include your GitHub profile name so I can share the repo with you.
-
Stock Watchlist
Tracks the stocks, ETFs and funds you're watching, and the ones you already hold across the family's portfolios. Set a target buy price and an alert level and rows light up when a stock enters the buy zone or crosses above it; macOS notifications also cover stop-loss and take-profit levels and an optional daily summary. Record buys, sales, dividends and cash movements, or import them from your bank's CSV export, to see average cost, realised and unrealised gain, worked out in DKK at each transaction's exchange rate. A portfolio summary charts value over time against a benchmark such as the S&P 500, tracks dividends and how much each aktiesparekonto can still take this year, and gives the year's figures per account as a CSV or printable report. Click any row for a price chart, fundamentals and recent news. Live data from Yahoo Finance. Runs locally on your Mac, optionally reachable with a password from phones on your home network, and stores everything in a single file you own, backed up daily.
Request access -
Millionærklubben Transcript & Panel Sentiment
Turns each episode of the Danish investing podcast Millionærklubben into data. Pulls the episode, transcribes it locally with Whisper on Apple Silicon, has Claude pick out the companies discussed, and appends tickers, panel sentiment and reported price moves to an Excel workbook. Run it from the terminal or a small local web UI that streams the log, shows the latest results, helps file mis-heard company names and follows your Mac's light or dark appearance. Follow one company across every episode — how often it was discussed versus just mentioned, its average panel score, and every mention playable from the episode audio to check a garbled name by ear. Scores are read as bands, not decimals. A personal research tool, not investment advice.
Request access -
Podcast Transcriber
Find a podcast by name, paste its RSS feed or a YouTube video link, or drop in a recording of your own, and it turns each episode into one Markdown file written for a chat model — and a 2–3 minute audio briefing for when you'd rather listen. The transcript has speaker turns as paragraphs, a labelled chapter every five minutes, and a header with the date, who's who, a short summary and topic tags, with pre-recorded sponsor spots cut to a one-line marker; an hour of talk is about 20k tokens. The briefing is a narrator on the key takeaways with the speakers in their own voices, playable on the page or delivered to your phone as a private podcast feed on the home network. Whisper large-v3 runs on the Apple GPU, pyannote separates the voices, and a local LLM (or Claude) names them and writes the summary — and measures its own naming accuracy against episodes you've corrected. Search any phrase across every transcript and play that moment of the audio, the text following word by word. New episodes can transcribe themselves hourly. Built for Danish, works for any language Whisper knows. Terminal, browser or Mac app. Everything runs on the Mac: the only traffic is fetching the episodes (and a video's metadata and audio from YouTube, when you add one), the podcast-name lookup, a one-time model download — and Claude only if you choose to give it a key. MIT licensed.
Request access