# The Reason AI Gives You Useless Answers (Fix This First) > [!question] Lingering Questions > - How much perishable context is optimal before diminishing or negative returns set in (context bloat degrading focus)? > - Where is the real line between "context engineering" (technical) and "context management" (information work) — is Forte's objection substantive or just rhetorical? > - Does heavy persistent context risk over-anchoring the AI to stale preferences over time? > [!tip] Article Ideas > - Map Forte's three-layer model (persistent/project/perishable) directly onto the vault's own PARA-adjacent structure and Claude custom instructions — a concrete "here's my master prompt" post. > [!link] Cross Reference > - [[Personal Knowledge Management MOC]] — PCM is presented explicitly as the AI-era successor to PKM > - [[AI & the Future of Knowledge Work MOC]] — reframes "prompt engineering" as the wrong lever entirely > [!quote] Vocabulary > - *personal context management (PCM)* — Forte's coinage for curating persistent, project, and perishable context so an AI system can act on your behalf > - *master prompt* — a stable block of persistent context (role, values, preferences) stored once in custom instructions rather than repeated per conversation ## Bib capsule Tiago Forte (Building a Second Brain) argues in a demo-driven video that generic AI output is a missing-context problem, not a bad-prompt problem, and proposes a three-layer model — persistent, project, perishable — for staging context before you ask. ## Author's thesis (what is this source about / what is asserted) The bottleneck on AI output quality has moved from prompt phrasing to context curation; Forte renames the mature practice of personal knowledge management "personal context management" and demonstrates, live, that the same request goes from generic to executable purely by adding three layers of staged context — no change to the prompt itself. ## Key arguments (≤3) 1. **The misdiagnosis is prompt wording; the real gap is context**: "the problem isn't in your prompt, it's in the context you're giving, or more specifically, not giving it." 2. **Context has three layers — persistent, project, perishable**: a stable master prompt, an active project folder (PARA maps directly onto this), and a last-mile brain-dump specific to one request. 3. **This reframes AI use as delegation to a system, with the same constraint as delegating to a person**: "the quality of your handoff" — and the corollary that curating context is really curating yourself, externalizing judgment so the AI can borrow it. ## Claim atoms ### Generic AI answers come from missing context, not bad prompts > [!quote] ID:rw_pending > "The problem isn't in your prompt. It's in the context you're giving, or more specifically, not giving it." Where/why: [00:00:00] — opening thesis; reframes the prompt-engineering obsession as the wrong lever. ### Context has three layers: persistent, project, perishable > [!quote] ID:rw_pending > "Persistent, project, perishable. Three Ps, three layers. Together, they're what we mean when we say context." Where/why: [00:02:12] — persistent = master prompt, project = PARA folder, perishable = last-mile nuance; an operational taxonomy that maps directly onto an existing PARA/Obsidian setup. ### Personal context management is delegation to a system > [!quote] ID:rw_pending > "PCM, personal context management, is delegation, but to a system instead of to a person. And the constraint is exactly the same, the quality of your handoff." Where/why: [00:13:08] — the bottleneck moves from output to clarity of intention; imports management/delegation skill wholesale into AI work. ### You curate yourself, externalizing judgment so AI can borrow it > [!quote] ID:rw_pending > "You have to externalize your judgment so the AI can borrow it. What you're really doing in personal context management, I believe, isn't curating data, it's curating yourself." Where/why: [00:13:41] — counters AI's homogenizing/flattening effect; the distinctiveness lever is explicit taste, values, opinions. ## Permanent candidates - **AI output quality is bounded by staged context, not prompt phrasing** — strong, portable claim directly applicable to the vault's own Readwise → Obsidian → Claude workflow. Promoted below. ## Processing checklist - [ ] Draft a personal master prompt (persistent layer) and store it in Claude custom instructions - [ ] Map one active PARA project folder + Obsidian notes as a reusable project-context bundle - [x] Connect into the relevant MOCs — [[3. Resources/MOCs/Personal Knowledge Management MOC]], [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] ## Metadata - Source: [[3. Resources/Literature Notes/The Reason AI Gives You Useless Answers (Fix This First)]] ## Links - [[3. Resources/MOCs/Personal Knowledge Management MOC]] — PCM is presented as PKM's direct successor once notes become AI inputs, not just future-you inputs - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — reframes context curation, not prompt wording, as the operative skill for AI-era knowledge work