Rational Optimization \w Elias Schlie

This month, I built AI agents, I moved, I fasted, and I hiked. In this podcast, I share the stories that I experienced and the insights I found.

YouTube version

My website: eliasschlie.com

Quick Stats 📊⏱️ Total focused work: 288 hours and 46 minutes🗓️ That’s an average of 9 hours and 57 minutes every single day — weekends included!
💾 Full breakdown publicly available here: eliasschlie.com

What I spent my time on 🧠

• 🚀 Deep dive into AI agents and language model tooling (esp. langchain + agents)

• 🧪 Built a custom MCP server to connect LLMs with Obsidian for persistent memory scaffolding

• 🧭 Personal reflection on direction, meaning, and intrinsic values (inspired by valuism)

• 💬 Developing systems for better conversations and group dynamics

• 📚 Listening to high-signal content during breaks to stay immersed in key problems

Main Achievements 🏆

✅ Built and shipped my own MCP server — clean, simple, and powerful

✅ Reached a new level of clarity on what I value and why it matters

✅ Used fasting as a powerful tool to reset and return to deep focus

✅ Refined my approach to learning through better break design and structured reflection

✅ Pushed physical + mental boundaries with a 50h fast + night hike — reminder of agency over mind and body

**June 2025: The Month I Started Building**

Quick Stats 📊

⏱️ Total focused work: \~300 hours

🗓️ That’s 9 hours and 49 minutes every day (including weekends)

💾 Full work log published at [https://github.com/EliasSchlie/Working\_Hours](https://github.com/EliasSchlie/Working_Hours)

What I spent my time on ⚡

* Finished exams for 9 university courses in Cognitive Science & AI (across three different semesters)

* Deep work on AI agents, automation, and building new tools

* Built foundational scripts for automating my note-taking (Obsidian) with LLMs and tools

* Explored and prototyped AI agents for science (especially agents to automate news & paper summarization)

* Researched and learned more about AI agent architectures and how to use tools/APIs for automation

* Started serious learning in JavaScript/TypeScript and web development (CS50 Web, JavaScript fundamentals course)

* Continued refining systems for both learning and building

Main Achievements 🏆

✅ Completed 9 courses in about a month, with top or near-top grades in multiple cohorts

✅ Built the first versions of AI agents to automate my note-taking and science workflows

✅ Prototyped a system to automatically create and publish longevity science podcast episodes from new research papers

✅ Developed and used structured note-taking, now over 10,000 notes (“second brain”)

✅ Shifted from pure learning mode to building mode, exploring new personal systems for productivity

✅ Stayed consistent with high work hours, while keeping balance with gym, sauna, and friends

✅ Maintained strong intrinsic motivation—doing exactly what I’d do even if given €1 million

⁠YouTube⁠ version

test2

eliasschlie.com

eliasschlie.com

My website

• 🐭 FDA Plans to Replace Animal Testing — A Roadmap to Reducing Animal Testing in Preclinical Safety Studies

• 🤖 Why Didn't AI Discover a Bunch of Drugs Yet? — Progress, Pitfalls, and Impact of AI-driven Clinical Trials (Wilczok & Zhavoronkov, 2025)

• ⏰ LinAge2 – Aging Clock Update — Providing actionable insights and benchmarking with epigenetic clocks (Fong et al., 2025)

• 🚀 Frontier Tower — Permanent Longevity Hub in San Francisco

• 🏝️ Vitalist Bay — Giant Longevity Pop-Up City

My website

1) 🐁 The Rodent Aging Interventions Database (RAID)

2) 🧫 High-Throughput Yeast Aging Study

3) 🤖 Metabolic Digital Twins

4) 🏙️ How Cities Can Be Used to Learn About Aging and Exposome Influences

5) 🐷 First-ever pig-human liver transplant

⁠⁠My Website

⁠My Website