PKM is not "capture and store." It is a multi-stage pipeline where raw information is progressively refined into actionable knowledge and creative output. The Knowledge Funnel makes this pipeline explicit: each stage has distinct activities, note types, tools, and review cadences. The funnel narrows deliberately. Many inputs enter; fewer, more refined outputs emerge.
The Stages
1. Discovery and Curation. Selecting what to pay attention to. RSS feeds, newsletters, book recommendations, conversations, social media. The key activity is filtering: most information should be ignored. See Information Diet.
2. Capture. Getting selected material into the system. Web clipping, voice memos, quick notes, photos of whiteboards. Speed and low friction matter here; judgment does not. See The Capture Habit.
3. Literature Processing. Reading, annotating, and summarizing captured sources. Highlights become literature notes. The goal is to extract what matters from the source in your own words. See Progressive Summarization.
4. Knowledge Synthesis (Dots). Generating sub-atomic fragments: observations, hunches, principles. This is the reflective stage where breadcrumbs, seeds, and wisdom emerge from processed literature and personal experience. See Dots and Sub-Atomic Knowledge.
5. Permanent Knowledge Creation. Promoting refined dots into atomic notes that stand alone, are written in your own words, and express a single idea. See Atomic Notes.
6. Connection. Linking permanent notes to each other, forming clusters, discovering unexpected relationships. This is where Compounding Knowledge begins.
7. Creation. Composing outputs (articles, presentations, decisions, products) from connected atomic notes. The notes become building blocks for original work. See Writing as Thinking.
8. Sharing. Publishing, teaching, discussing. Output returns to the wider information ecosystem, potentially entering someone else's funnel at stage 1.
The Narrowing Principle
The funnel narrows at every stage. You discover hundreds of sources but capture dozens. You process dozens of captures but synthesize a handful of dots. You promote a few dots to permanent notes. You compose one article from many notes. This narrowing is not waste; it is the value creation mechanism. Each stage concentrates signal and discards noise.
Stage-Specific Cadences
Each stage operates on a different rhythm. Capture is continuous (throughout the day). Literature processing is batched (weekly reading sessions). Synthesis happens during periodic reviews. Connection is ongoing but deepens during dedicated thinking sessions. Creation follows project timelines. Recognizing these natural cadences prevents the common mistake of trying to do everything in a single sitting.
Relationship to Knowledge Lifecycle
The Knowledge Funnel overlaps with the Knowledge Lifecycle but is more concrete and workflow-oriented. The Lifecycle describes the full arc including maintenance, review, and pruning. The Funnel describes the forward pipeline from raw input to refined output. Together they cover both the creation path and the maintenance path.
Key Points
- The Knowledge Funnel is an eight-stage pipeline from discovery to sharing
- Each stage has distinct note types, tools, and cadences
- The funnel deliberately narrows: many inputs, few refined outputs
- The narrowing is the value creation mechanism, not information loss
- Complements the Knowledge Lifecycle, which covers the maintenance arc
Open Questions
- What is the typical conversion rate at each funnel stage (e.g., what percentage of captures become permanent notes)?
- Can AI accelerate the middle stages (literature processing, synthesis) without undermining comprehension?
- Should the funnel be strictly sequential, or can stages overlap and loop?
References
- Vault: Obsidian Starter Kit - Theory, Personal Knowledge Management Process, Idea Development Environment (IDE)
- Tiago Forte, "Building a Second Brain" (2022) — CODE framework as a simplified funnel
- Sönke Ahrens, "How to Take Smart Notes" (2017) — on the progression from fleeting to permanent notes