How I Built an AI Research Skill That Checks Its Own Citations
The author created an AI research skill called Jean, built on Claude Code, that adds structured scoping, three‑stage literature searches, and automated citation verification. By pulling from academic databases such as Semantic Scholar and arXiv, then expanding via author and citation trails, the skill ensures every reference includes a verifiable DOI or URL. Hallucinated citations drop dramatically—GPT‑3.5 fabricated 55% of references, while GPT‑4 still produced 18%, a gap Jean’s verification loop eliminates. The result is a reliable, ready‑to‑read literature review that saves hours of manual fact‑checking.
You're Invited to Spaced Reps, a Bi-Weekly Meeting to Level Up Your Anki Deck
Starting May 26, Eva Keiffenheim launches a bi‑weekly online session called “Spaced Reps” aimed at helping Anki users create, prune, and refine their flashcard decks. The meetings are positioned as a paid‑subscriber benefit, offering live expert guidance and peer feedback. Participants...
I Used to Say "I Have a Bad Memory" But Now I Know It Was Just Untrained
The author discovers that a "bad memory" is often a symptom of untrained recall skills rather than a fixed flaw. Interviews with six‑time USA Memory Champion Nelson Dellis and a 2017 Neuron study show that systematic mnemonic training can double...
When AI Flashcards Pollute Your Anki Deck (And the 4-Step Workflow That Makes Them Useful Instead)
AI tools that auto‑generate Anki flashcards promise speed but often deliver noisy, ambiguous cards. A 2026 benchmark found that even top‑tier models like GPT‑5.2 produce unusable cards 36% of the time, eroding deck quality. Cognitive research shows self‑generated cards improve...
Why the Smartest Choice Might Be to Ignore the Shortcut
The post warns that AI’s confident, fluent answers can lull users into uncritical reliance, echoing the author’s experience with an over‑confident mentor. It highlights three hidden harms: hidden biases in training data, cognitive offloading that weakens critical thinking, and the...
How to Keep Your Anki Deck Relevant and Useful
The author admits abandoning his own spaced‑repetition habit and explains why AI’s rise and deck bloat made his Anki cards less useful. He argues that memorization remains vital, but only for information that supports critical thinking and real‑world decisions. The...
4 Attentional States That Explain Why You Reach For Your Phone 47 Seconds Into A Book
Gloria Mark’s two‑decade research shows the average screen‑switching time for knowledge workers fell from 2.5 minutes in 2004 to about 47 seconds by 2020. Crucially, the data reveal that 44% of those switches are self‑initiated, meaning we often distract ourselves...
How to Delegate to AI Without Lowering Your Standards
AI product manager Karo Zieminski and researcher Eva Keiffenheim discuss how they delegate tasks to Claude Cowork without compromising quality. They emphasize framing prompts around the desired outcome, feeding the model finished work for repurposing, and using it to handle...
A Short, Yet Useful Guide to Claude Cowork
Claude Cowork, Anthropic’s new desktop‑only AI agent, lets users assign concrete outcomes to folders of files rather than asking questions. By reading PDFs, notes, or Obsidian vaults, it plans, executes, and writes deliverables such as briefs, spreadsheets, or slide decks...
The Most Important AI Skill Has Little to Do With Prompting (And A Lot to Do With How Humans Learn)
The article argues that the most valuable AI capability isn’t clever prompting but the ability to structure interactions as reusable "Skills," akin to an employee following a standard operating procedure. Choosing the right model—Claude Opus for deep reasoning or Gemini...