// training paths · six tracks, in order
Training paths
Each pillar as an ordered track. Start at the top of any track; j/k steps you through once you're inside a module.
0/76 logged · no uptime yet — the first module starts the clock
·
Directing the model; what you feed it
2026·06·08 Brief the AI like a brilliant new hire: writing clear, direct instructions 2026·06·08 Context engineering: curate the few high-signal tokens, don't dump everything in 2026·06·08 Beyond the prompt recipe: iterative collaboration, not a better template 2026·06·18 Does 'Act as a senior lawyer' actually help? The truth about role prompting 2026·07·06 Say the what, delegate the how: specify the outcome, not the recipe 2026·07·21 The density tax: why a short, fact-packed paste can be harder than a long one 2026·07·27 Front-load the goal: why a late correction can't save a long AI task 2026·07·29 Personas don't transfer: the role prompt that helps on one model can hurt on the next 2026·08·04 It noticed your question couldn't be answered. Then it answered. 2026·08·06 Ask for the edit, not the rewrite 2026·08·20 Make each rule a yes or no: why half your saved instructions stopped firing 2026·08·31 The rule you wrote and the rule that binds 2026·09·01 Flatten the sheet before you ask: the layout is what it gets wrong, not the numbers 2026·09·09 Ask again in a clean thread 2026·09·21 Ask for three, then pick: you can't describe what you haven't seen
What persists across turns and sessions; retrieval
2026·06·08 Your context window is working memory, not a hard drive 2026·06·08 RAG without the jargon: how AI can answer from your documents instead of guessing 2026·06·16 Own the memory, not just the model: the harness layer that quietly traps you 2026·06·25 It only sees your slice: how permissions shape what your work assistant tells you 2026·07·08 Connected isn't constant: when your assistant actually reaches into your inbox 2026·08·12 Your saved summary is an index, not evidence 2026·08·13 The hedge doesn't get saved
Extending the model: integrations, connectors, setup
2026·06·08 Connectors vs. file upload: when to plug into your live data and when to just drag a file in 2026·06·11 Projects, Custom GPTs, and Gems: which reusable workspace to build 2026·06·23 Citation theatre: the more a research agent cites, the more links it fabricates 2026·06·30 Be your own connector: feeding a sandboxed AI what it can't reach 2026·07·14 Show, don't tell: paste the screenshot instead of describing it 2026·07·22 The Silent Engine Swap: What to Do When Your AI Tool Changes Underneath You 2026·08·03 Hidden instructions in the document: a file you didn't write is untrusted input 2026·08·05 The model picker is a data control, not a quality dial 2026·08·10 A citation isn't proof: check what your assistant actually used 2026·08·19 Effort is a trade, not a dial: when to turn the thinking down 2026·09·03 Which transcript did you get? 2026·09·15 The more it retrieved, the less it asked 2026·09·16 It cited the old version
Multi-step work, orchestration, refining to a result
2026·06·08 The generate-critique-refine loop: stop accepting first drafts 2026·06·08 Plan first, execute second: the 80/20 rule for letting AI tackle big tasks 2026·06·17 Briefing a Deep Research Agent: How to Get a 20-Source Report Worth Reading 2026·06·29 Convene a Panel: Using Multiple AI Personas to Stress-Test a Decision 2026·07·01 The handoff note: let the outgoing session write it for you 2026·07·13 The pre-mortem: assume the plan already failed, then ask why 2026·07·16 Read the why, not just the fix: using AI's formula repair as a free lesson 2026·08·11 When your AI chat gets lost, don't patch it — restart it clean 2026·08·17 Write the bad first draft yourself 2026·08·26 It worked when you watched it: what a single success actually buys 2026·09·08 Give it a way out 2026·09·14 "How did I do?" is not feedback
Knowing when not to; failure modes, over-reliance
2026·06·08 Why it makes things up: hallucination is a feature of the scoring, not a glitch you can prompt away 2026·06·15 Citation roulette: the verification habit that keeps you out of the AI-hallucination court database 2026·06·22 The account is not the artifact: judging AI work you didn't watch get made 2026·06·24 The summary that ate the caveat: what AI quietly drops when it condenses your documents 2026·07·02 It agrees because you pushed back: spotting sycophancy before it changes your mind 2026·07·09 The give-up reflex: what ten minutes of AI help does to your persistence 2026·07·15 Verify the request, not the voice: why spotting the deepfake is the wrong defence 2026·07·23 Inherited capture: on someone else's call, the note-taker runs on their defaults 2026·07·28 The verification divide: why the same AI helps the best and hurts the rest 2026·07·30 A share link is a public URL: what the 'share chat' button actually does 2026·08·18 Ask for a hint, not the answer: taking the help without losing the skill 2026·08·24 The harsh verdict gets a pass 2026·08·25 Judge first, then ask: the order that beats double-checking 2026·09·07 What's missing doesn't look wrong 2026·09·10 It copied without the quote marks 2026·09·17 Leave out who said it
Reading model developments; signal from hype
2026·06·08 The jagged frontier: why AI aces one task and bombs the next one that looks identical 2026·06·12 Leading through the dip: the productivity J-curve and the real work of the AI transition weekly 2026·06·12 Colleague or contractor: the two ways to hand work to an AI 2026·06·19 The rented mind: when intelligence is cheap, memory is the lock-in weekly 2026·06·26 The imagination gap: when execution is cheap, judgement is the divide weekly 2026·07·03 The supervision paradox: why more capable AI means more babysitting, not less weekly 2026·07·07 The tool is not the model: what your AI can do keeps changing under you 2026·07·10 Vibe citing: when generation is free, being able to vouch for it is the job weekly 2026·07·17 The ambition technology: why maximum AI value starts with a bigger ask, not a smaller workload weekly 2026·07·20 Why the Leaderboard Lies: Reading AI Benchmarks Without Getting Fooled 2026·07·24 Measurement debt: when belief in AI outruns the proof weekly 2026·08·27 A flag is not a verdict 2026·09·02 Who was counted
// progress is stored in this browser only — nothing leaves the machine.
// wipe it any time: press ~ and run `log clear`.