Changelog
A running log of what I'm building, thinking, shipping, and learning.
Spent a session reworking the ThroughlineIcon animation. The octahedron’s spin axis now tilts and slowly precesses, so you see every face over time. Added acceleration that ramps from 1x to 3x with an easeOut curve, a bigger intro pulse, and a snappier collapse.
Installed a Google MCP server that connects Claude Code directly to Gmail, Google Calendar, and Drive. Read emails, listed this week’s events, and created a calendar invite with an attendee — all from the terminal.
Launched yofriedman.com. Built with Astro, deployed on Vercel. Design system, four pages, content collections, dark mode. The site is a living changelog drafted by an agent and edited by me, capturing what I’m building with AI and why.
Refactored the personal vault. Created a Massachusetts AI Coalition project space, consolidated standalone files under one roof. I like how the vault keeps getting simpler as the system gets smarter.
Built a /log skill and updated /handoff to draft changelog entries automatically. Handoff proposes, /log reviews and publishes.
Moved all scripts into the vault repo and set up symlinks for cross-machine sync. ~/scripts/ now points to ~/vaults/throughline/scripts/ — git pull on any machine gets the latest automation. One less thing to maintain manually.
The valuable thing isn’t “AI enablement.” Nobody wakes up wanting to be AI-enabled. They want their invoices processed, their fleet tracked, their data visible. AI is the delivery mechanism. That reframe changed how I talk about what I do.
Delivered a statement at the Massachusetts AI Coalition kickoff as the co-lead of the Banners task force. The framing: this is not just a technology revolution, it is a human revolution. Technology friction is temporary. The enduring challenge is helping people unlock and build without fear.
The moat isn’t technical. Tools get easier. The moat is knowing what to build (domain judgment no tool provides), the human transfer (making someone believe they can do this), and carrying the message (writing, speaking, inspiring).
Refactored complex workflows to use an orchestrator and sub-agent pattern. Heavy research gets dispatched to background agents with their own context windows. They return compressed briefs to the main conversation. The primary context window stays clean while doing deep work. This is the pattern for every future complex skill.
Added Chrome DevTools MCP to the Throughline workflow — Claude Code can now drive an authenticated browser session. Google Docs, Slides, dashboards, anything behind a login. The gap between “I need a document” and “here’s the document” is now a single command.
Installed Firecrawl for scraping web pages to clean markdown. Any URL becomes a searchable document in the vault with one command. The gap between “I read something interesting” and “it’s in my system” is now about five seconds.
Built a two-layer continuity system for AI sessions. Shell hooks fire automatically on start and end. A handoff command captures rich context to the project file. The insight: context continuity is the hardest problem in AI-augmented knowledge work. The solution is behavior design, not just tooling. Make the right thing the easy thing.
Migrated task management from markdown files to Linear via MCP. Two systems, clear roles: Linear holds what to do and when. The vault holds why and what happened. Five slash commands bridge the two. Tasks in a real tool, context in owned files.
Added QMD, a local semantic search engine over the entire vault. 1,300+ markdown documents, searchable by meaning, not just keywords. Models run locally. Claude can now find relevant context across years of notes and meeting transcripts in under a second.
Built a pre-meeting agenda command. It scans all project files for stakeholder matches, reads past meetings with that person, and synthesizes a smart agenda. The prep work I used to spend 15 minutes on now happens in the time it takes to type /agenda.
Discovered that AppleScript doesn’t properly expand recurring calendar events. Rewrote the daily sync script using icalBuddy instead. Third rewrite of this script. The lesson: the first version of any automation is wrong. The value is in the iteration, not the initial build.
Documentation is not an artifact for onboarding. It is source code for knowledge work. In an era of capable AI, the opportunity cost of not having properly engineered context is enormous. This is the idea I keep coming back to.
Restructured how meeting notes work. The daily file became an index with brief summaries and links, not a repository. Detailed notes route by type: recurring meetings get their own files, project meetings go to project files. The system scales now because _Daily.md stays lean and context lives where it’s useful.
Wrote a letter to a friend explaining the system I built and why she should build her own. The line that stuck: “English is a programming language now.”
The constraint in knowledge work is shifting from execution to context. AI can write a memo. It cannot write your memo without your institutional memory. The people who get the most leverage will have rich institutional memory, in a form AI can read, that they own.
Onboarded two people outside my company onto their own AI operating systems. Built a repeatable three-document framework: a vision spec, a setup guide, and an interactive onboarding doc that Claude executes like a program. Each onboarding took about 10 hours and produced a working system for a completely different business.