Michael Saylor Knowledge Base — Definition of Done

v1.0 — 2026-03-03

Context

Michael Saylor has done thousands of hours of public speaking — podcasts, news interviews, conference keynotes, corporate earnings calls, and Bitcoin-focused presentations. He's one of the most quotable, framework-heavy thinkers in the Bitcoin space, but his knowledge is scattered across hundreds of YouTube videos, podcast feeds, and corporate archives. You can't search it, cross-reference it, or ask it a question.

What makes Saylor unusual isn't just the volume — it's the discipline. When asked about Dogecoin, he stays in his lane. When markets crash, he's cheerful and constructive. He's built a library of mental models and repeatable frameworks for thinking about money, energy, and civilization — and he deploys them with laser focus on Bitcoin. That signal-to-noise ratio is what makes this corpus worth building into a knowledge base.

I want to scrape every Michael Saylor interview and public appearance from the internet and transform it into a semantic search knowledge base — fully transcribed, speaker-identified, categorized, and queryable by natural language.

The Outcome

I ask "How would Saylor respond to someone who says Bitcoin is too volatile to be a treasury asset?" — and I get his actual answer, synthesized from dozens of appearances where he addressed that exact objection, with his signature analogies intact. Cheerful, constructive, laser-focused on Bitcoin. Sourced, dated, and weighted toward his most recent thinking.

It feels like having Saylor in the room — his discipline, his mental models, his refusal to get pulled off-topic — except he's searchable, organized, and you can cross-reference what he said on Lex Fridman against what he told the SEC.

The Experience

I'm preparing for a debate about Bitcoin as a corporate treasury strategy. I ask the KB: "What's Saylor's steel-man case for why Bitcoin beats bonds, gold, and real estate as a store of value?" Within a minute I get his full framework — the thermodynamic argument, the digital energy analogy, the comparison to Manhattan real estate — sourced from 8 different appearances across 3 years, with the most recent arguments weighted highest. Each claim links back to the specific interview and timestamp.

A teammate asks: "What analogies does Saylor use to explain Bitcoin to normies?" They get a ranked list of his most-used analogies — digital gold, cyber Manhattan, melting ice cube, mosquito net over your wealth — with the exact quotes and which interviews he deployed them in.

Done When

1. Every public appearance transcribed & speaker-labeled

• All discoverable Saylor interviews, podcasts, conference talks, and corporate presentations scraped and transcribed
• Every speaker identified by name (Saylor, interviewer, panelists, analysts)
• Saylor always tagged as primary speaker
• Test: Pick any 3 random files → each has a clean, speaker-labeled transcript with no unresolved speaker tags

2. Categorized by appearance type

• Every transcript tagged by format: podcast, news, conference, corporate
• Can filter by category or search across all
• Podcasts tagged with show name (Lex Fridman, What Bitcoin Did, etc.)
• Corporate presentations tagged by type (earnings call, investor day, Bitcoin strategy update)
• Test: "Show me only his conference keynotes from 2024-2025" → returns only conference content, cleanly tagged

3. All participants identified & tracked

• Every interviewer, host, panelist, and analyst identified by name across the entire corpus
• Can query by person: "What did Lex Fridman ask Saylor about?" or "Which CNBC anchors have interviewed him?"
• Tracks who challenged him hardest, who gave him the most room to monologue, recurring interviewers
• Test: "Who has interviewed Saylor more than 3 times?" → returns names with counts and dates

4. Knowledge distilled into structured frameworks

• Thinking frameworks & mental models — how he thinks (thermodynamic store of value, digital transformation of X, 100-year time horizon, "cheerful & constructive" as a deliberate operating stance, staying in your lane, laser focus as competitive advantage)
• Playbooks & SOPs — his repeatable processes (how to pitch Bitcoin to a board, how to structure convertible note offerings, how to respond to regulatory FUD)
• Systems & leadership — his organizational principles (MicroStrategy's Bitcoin strategy execution, corporate governance around BTC treasury)
• Philosophy & money frameworks — his foundational beliefs (why fiat fails, what money is, Austrian economics meets digital age, energy-money equivalence)
• Each framework sourced to specific appearances with quotes
• Test: "What's Saylor's framework for evaluating whether an asset is money?" → returns his full hierarchy (scarcity, durability, portability, divisibility, programmability) with sources

5. Timestamped & recency-weighted

• Every piece of content has a date
• Queries weighted by recency by default (his 2025-2026 thinking on ETFs > his 2020 thinking)
• Can also search for timeless frameworks without recency bias
• Test: "What does Saylor say about Bitcoin ETFs?" → 2025-2026 weighted higher. "What is digital energy?" → best explanation regardless of date

6. Saylor-voice query responses with analogies

• Can ask "How would Saylor respond to X?" and get an answer synthesized in his voice — using his actual analogies, not generic summaries
• Analogies catalogued and tagged (which ones he uses for which arguments)
• Cross-references multiple appearances to build the strongest version of his argument
• Test: "How would Saylor convince a traditional CFO to add Bitcoin to the balance sheet?" → response uses his actual analogies (melting ice cube, stranded on an island with $100M), cites specific interviews, reads like Saylor wrote it

7. Semantic search with citations

• All content embedded and queryable via natural language
• Every answer cites specific interviews with dates and speaker context
• Can ask esoteric cross-cutting queries ("How has Saylor's view on altcoins evolved from 2020 to 2025?")
• Test: Ask 5 real questions → 4/5 return useful, sourced answers with specific interview citations

8. Automated ingestion pipeline

• New Saylor appearances detected and ingested automatically (YouTube channels, podcast feeds, MicroStrategy IR page)
• Full pipeline: discovery → download → transcription → speaker labeling → categorization → framework extraction → embedding
• Pipeline failures alert rather than silently skip
• Test: Saylor does a new podcast interview. Within 48 hours it's transcribed, labeled, categorized, and searchable — no manual intervention

Stretch Goal: Reusable across public figures

• The entire pipeline can be re-pointed at a different public figure (Balaji, Naval, Buffett) without rebuilding from scratch
• Configure: source channels/feeds, speaker name, category taxonomy — pipeline handles the rest
• Test: Re-point the pipeline at Mike Israetel (bodybuilding / Renaissance Periodization). Retool the category taxonomy, configure source channels, and deploy. The system produces a working, queryable KB for a completely different person in a completely different field — without rebuilding from scratch

Feedback Form

Not Done Until Brad Says So

Brad tests with 5 real questions he actually cares about. 4/5 return useful, sourced, Saylor-voice answers — it ships.

Not In Scope

• Private conversations or leaked content
• AI-generated "what Saylor would say" beyond what's in the actual corpus
• Web UI or app (future phase)
• Real-time monitoring of live appearances

Changelog