# Arthur Pfalzgraf > Personal site of Arthur Pfalzgraf, Director of Product at Iron by MoonPay (stablecoin payments). Previously product at Ultimate (acquired by Jupiter Exchange), Trade Republic and HelloFresh. The site logs what he builds on the side: AI agents, n8n automations, and the data layer underneath them. ## Pages - [About](https://arthurpfz.com/experience/): career history and short bio - [Builds](https://arthurpfz.com/builds/): curated list of personal projects with stack and links - [Full text of the articles](https://arthurpfz.com/llms-full.txt): build write-ups as plain markdown ## Build write-ups - [Croissant Ride](https://arthurpfz.com/archive/croissant-ride/): My own indoor training app, cut to the essentials. Runs in the browser, open source. - [A tri-coach on Telegram](https://arthurpfz.com/archive/the-coach-that-watches-the-data/): An AI triathlon coach that grades every workout against your fitness, writes next week's plan on its own, and lives entirely on infrastructure I control. The architecture, the components, and the three patterns that hold it together. - [A vault that ingests itself](https://arthurpfz.com/archive/stop-typing-second-brain/): Voice notes, bookmarks, tweets, YouTube — captured automatically, filed in markdown, queryable from anywhere. The architecture, the schema, and the three patterns that hold it together. ## Other builds - Daily Podcast — my morning briefing, read to me: A two-host, ~15-minute podcast generated every morning: overnight macro / crypto / AI news, prediction markets, Hacker News, yesterday's bookmarks, and one training tip from my vault. Lands on Telegram around 06:00. About $0.10 an episode. - Telegram capture bots: Two Telegram bots for getting things out of my head and inbox: send a tweet, article or YouTube link and get a clean markdown note in my vault; or record a voice memo and get it transcribed and filed. Same surface, two jobs. - Brain Digest — one vault page, every Monday: Every Monday at 08:00, one recently updated page from my vault comes back to me as a ~150-word Telegram dispatch: one tension, gap or surprising fact, not a summary. A digest fails if it could have been written a month ago. One n8n workflow. - Voice notes → personal wiki: Telegram voice memo → Whisper transcription → Claude categorisation (idea / todo / other) → markdown file written into my Obsidian vault, with an undo button. Whole pipeline is one n8n workflow. - Hermes — an agent that lived in my vault: VPS-side agent with two personalities: 'vault-aware' answers questions over my knowledge base in read-only mode, 'vault-writer' drafts new pages into a sandboxed inbox folder. Talks to me on Telegram. Ran on Grok via OpenRouter. Retired in November 2026: it had ended up only running fixed scripts on a schedule, and that job belongs in n8n. - Relay — running my VPS from Telegram: Webhook-based command runner so a phone-side Claude (or just me, on a train) can restart a container, tail logs, redeploy, or run an arbitrary script on the box. Auth + audit trail + a small allow-list of operations. n8n is the dispatcher. ## Book summaries - [Designing for behavior change, S. Wendel](https://arthurpfz.com/archive/designing-for-behavior-change/): Notes on Stephen Wendel's Designing for Behavior Change: how the mind decides, picking the right target behavior, and designing and refining a product that helps people act. - [Black Box Thinking, M. Syed](https://arthurpfz.com/archive/black-box-thinking/): Notes on Matthew Syed's Black Box Thinking: why aviation learns from failure and medicine hides it, how cognitive dissonance blocks learning, and what a culture of testing looks like. - [Good Strategy Bad Strategy, R.Rumelt](https://arthurpfz.com/archive/good-bad-strategy/): Notes on Richard Rumelt's Good Strategy Bad Strategy: the kernel of a good strategy (diagnosis, guiding policy, coherent action) and the hallmarks of a bad one: fluff, goals and wish lists. - [Cracked It!, O.Sibony](https://arthurpfz.com/archive/cracked-it/): Notes on Cracked It! by Garrette, Phelps and Sibony: solving big problems like a strategy consultant, the five pitfalls of problem-solving, and how to sell the solution. - [Hooked, N.Eyal](https://arthurpfz.com/archive/hooked/): Notes on Nir Eyal's Hooked: the Hook Model for habit-forming products. Trigger, action, variable reward, investment, with examples from Amazon, Facebook and Evernote. - [Predictably Irrational, D.Ariely](https://arthurpfz.com/archive/predictably-irrational/): Notes on Dan Ariely's Predictably Irrational: the decoy effect, anchor prices, the pull of free, and the other systematic ways we fail to decide rationally. - [Don't Make Me Think, S.Krug](https://arthurpfz.com/archive/dont-make-me-think/): Notes on Steve Krug's Don't Make Me Think: users scan, satisfice and muddle through, so make every page self-evident. The core rules of web usability. - [Brainfluence, R.Dooley](https://arthurpfz.com/archive/brainfluence/): Notes on Roger Dooley's Brainfluence: neuromarketing in practice. The pain of paying, anchors and decoys, precise pricing, and how the unconscious brain drives buying decisions. - [Speaking Up Without Freaking Out, M.Abrahams](https://arthurpfz.com/archive/speak-without-freaking/): Notes on Matt Abrahams' Speaking Up Without Freaking Out: where speaking anxiety comes from and practical techniques to manage it before and during a talk. - [Inspired, M.Cagan](https://arthurpfz.com/archive/inspired-marty-cagan/): Notes on Marty Cagan's Inspired: what separates good product teams from the rest. Outcomes over output, continuous discovery, and engineers involved from the first prototype. - [Outcomes Over Outputs, J.Seiden](https://arthurpfz.com/archive/outcomes-over-outputs/): Notes on Josh Seiden's Outcomes Over Outputs: an outcome is a change in customer behavior that drives business results. How to find, plan around and measure outcomes. - [The Lean Product Playbook, D.Olsen](https://arthurpfz.com/archive/lean-product-playbook/): Notes on Dan Olsen's The Lean Product Playbook: the product-market fit pyramid, problem space vs solution space, the Kano model, and scoping an MVP. - [User Stories Mapping, J.Patton](https://arthurpfz.com/archive/user-stories/): Notes on Jeff Patton's User Story Mapping: stories as shared understanding rather than requirements, mapping the big picture, and planning to build less.