[[Graphify]] extracts a confidence-tiered knowledge graph from documents — for [[Personal Knowledge management|personal knowledge management]] the question is whether that extraction adds enough discovery value over a hand-authored vault to justify the [[LLM]] tokens. ## [[Graphify]] ![[Graphify#^definition]] Scoped to the documents side at personal-corpus scale - prose, PDFs, articles, notes; the code pass is set aside. ## [[Personal Knowledge management]] fit For personal knowledge management the relevant requirements are the four baseline axes from [[Knowledge management]] - the three collective-only axes (Securability, Ownership clarity, Agent-readiness for multi-author governance) drop out because the single user is the producer, reviewer, and consumer. ### [[Discoverability]] ![[Knowledge problems#^discoverability]] Strong. The god-nodes report and graph traversal surface concepts that a flat folder leaves invisible. At personal-corpus scale, the legibility issue rarely bites - most users will not reach the size where god-node neighbourhoods become unreadable. Findability holds where it matters. ^discoverability ### [[Maintainability]] ![[Knowledge problems#^maintainability]] Mixed. Incremental updates skip unchanged files via the SHA256 cache, cheap when one document changes. But cross-batch semantic similarity drifts as new documents arrive; periodic full rebuilds restore it at the cost of an LLM pass over the whole corpus. At personal scale this is tolerable - the corpus changes slowly enough that occasional rebuilds are not disruptive. ^maintainability ### [[Connectivity]] ![[Knowledge problems#^connectivity]] Mixed. Edges across PDFs, transcripts, prose, and images land in one graph without prior schema work. But edges carry confidence tiers - they are claims, not facts. For PKM the multi-author audit concern disappears, but the human-as-author concern remains: connections are LLM-extracted, not authored, and the user cannot add or annotate edges directly in Graphify's output. ^connectivity ### [[Survivability]] ![[Knowledge problems#^survivability]] Weak. Source documents survive in the editing tool; the graph itself is derived, and its schema is emergent. A rebuild on the same corpus may produce a different relation vocabulary. The artefact survives the tool only in shape, not in detail. ^survivability ## Verdict - [x] [[#Discoverability]] - [/] [[#Maintainability]] - [/] [[#Connectivity]] - [-] [[#Survivability]] ^verdict Adopt as a secondary discovery layer behind your editing tool - useful for finding connections across a personal documents collection (PDFs, articles, notes, transcripts). Do not adopt as the primary store; the graph is not reproducible in detail and the human cannot author edges directly. ^answer The conclusion lives in none of the requirements on its own. Discoverability is the personal-knowledge value Graphify adds; maintainability and connectivity are tolerable but compromised; survivability is the unresolved cost. For PKM the verdict is *useful behind your real tool, not as it*.