About

What this site is

A public record. Since 25 November 2022 the intention has been to write one dated observation a day, and to keep the longer work in view alongside it: essays, research projects with their sources, dated forecasts, and the products and experiments that came out of it.

Nothing here is backfilled. A day without an entry stays empty on the calendar. A forecast is never edited after it is made; it is assessed later, separately, with evidence. Statistics on the homepage are calculated from the files in this repository at build time.

Who

Geart Ferhati. Born in Tirana, Albania. Self-taught programmer since the age of eleven.

• Founded I-VOTE at 15 (2022–2023): a digital platform for student-government elections, the first of its kind in South-East Europe. It reached 30,000 users across 85 schools within eight months and was acquired by the Council of Europe in 2023.
• Co-founded Suapr at 17 (2023–2025): an AI-powered learning platform with 57,000 users across three countries and a team of eight. Its research question, whether learning can be made as engaging as social media, secured €35,000 in government research funding.
• Worked as AI Automation Engineer at UVC Partners, a B2B venture-capital fund in Munich, building the AI agents and data infrastructure behind the fund's Value Creation and investment teams.
• Studies Management & Technology (computer-engineering focus) at the Technical University of Munich.
• Entrepreneur in Residence at Entrepreneurs First (The Bridge), San Francisco.

Speaking and civic work include a panel at the World Forum for Democracy (Council of Europe, Strasbourg, 2024), a keynote on AI in education at Google DevFest Tirana (2024), and serving as Vice President and Head of the Digitalization Commission of the Albanian Youth Parliament (since 2023).

How to read the archive

• Daily Record is the spine: short, dated, public. Start at /daily.
• Thinking holds essays. Research holds long-form investigations with explicit source lists. Predictions holds dated forecasts and their later resolutions. Work lists products and experiments with a plain status.
• Everything is plain HTML first. Feeds: RSS, Atom. Machine-readable map: llms.txt.