Frameworks
From Build Your AI-Powered Second Brain
The five principles
Capture what matters, ignore the rest. Write for retrieval. One file per subject. Own the storage. Compound the value.
The capture test
Capture only if you will act on it, you will reuse it, or it changes how AI should help you.
The four knowledge types
Identity (slow-changing). Project (weekly). Reference (rare). Decision (accumulates).
The structure and the core move
identity.md, preferences.md, decisions.md, plus projects/, reference/, people/, and archive/ folders. The core move: load context first, then give the task. Default load is identity.md plus preferences.md; add the project file, decision log, or a people file only when the task needs it.
From Communication & Influence
The Signal Test
Stakes: does anything change for the reader? Specificity: could someone verify or disagree? Stance: do I know what the sender thinks and wants? Signature: could only this person have sent it? Score each 0 to 1, for a total out of 4.
The three-minute signal pass
Stakes in line one. One abstraction replaced with one specific. A stance the message commits to. One line only you could write. Cut a fifth, read it aloud, send.
The AI draft workflow
Brief first (stakes, position, specifics, reader, voice). Pressure-test the draft (signal check, skeptic, compressor, recipient simulation). Fact-gate every specific claim. Do an ownership pass aloud. Disclose AI assistance when the message is substantive and consequential.
The influence layer
Map the decider, the key influencer, the affected party, and the skeptic. Pre-wire the skeptic before a group sees the recommendation. Track every commitment a message makes and close the loop on each one.
From Pressure-Test My Strategy
The seven weaknesses
Buried assumption. Chosen baseline. Hockey stick. Friendly pilot. Missing alternative. Convenient source. All-or-nothing ask.
The stress test
Strip the recommendation to its claim stack. Rate each assumption on load-bearing and uncertainty. Attack the evidence (source, age, denominator, counterfactual, survivors, reference class). Compare against steelman alternatives. Break it with a pre-mortem, then set tripwires and responses. Rebuild.
The personas and the answer architecture
Verifier, capital allocator, operator, historian, and where the room needs them, a regulator and an affected party. Answers follow a fixed shape: the direct answer first, the evidence second, the implication for the decision third.
The pre-presentation protocol
T-minus 7: stress test and rebuild. T-minus 4: question bank and answers. T-minus 3: pre-read check. T-minus 2: murder board. T-minus 1: one-pager and proactive disclosure. Day of: nothing new.
Prompt library
A working subset. The full appendix for each course has more, organized by workflow stage.
From Build Your AI-Powered Second Brain
The context frame
The files below are my personal context. Read them
before responding. Use them to match my voice, respect
my constraints, and reference my actual projects and
people. If you need a fact they do not contain, say so
rather than inventing one.The extraction prompt
Extract from these notes, in three short bullet lists:
1. Decisions made (what, by whom, stated reason)
2. Action items (what, owner, due date)
3. Open questions (what is unresolved, who answers it)
Use full names and roles where given. Write "not stated"
rather than guessing.The weekly review
[identity.md + active project files + recent decisions]
1. Which status lines are stale or contradictory?
2. Which open questions have sat longest?
3. Which decisions this week are not yet logged?
4. Given my priorities, what deserves attention next week?From Communication & Influence
The full brief
Draft [message type] to [audience].
The stakes: [why this matters to them, what it affects,
what happens if they do nothing]
My position: [what I recommend or have decided, and why]
Must include: [the specific facts: numbers, dates, names]
This reader: [what they care about, what they pushed
back on before, how formal we are with each other]
My voice: [how you write, described concretely, or
paste your identity file]
Put the stakes and the ask in the first four lines.
Do not add facts I have not given you. Where you need
a fact I have not supplied, insert [NEEDED: description]
instead of inventing one.The signal check
Score this draft against four criteria:
1. Stakes: does it state consequences for the reader?
2. Specificity: does it make claims someone could
verify or dispute?
3. Stance: does it take a clear position?
4. Signature: does it sound like a specific person
rather than anyone?
Identify the weakest of the four and show me exactly
which sentences fail it.The fact gate
List every specific claim in this draft: numbers,
dates, names, quotes, and statements about what other
people said or decided. Present them as a checklist
for me to verify. Mark any that you generated or
inferred rather than took from my input.From Pressure-Test My Strategy
The seven-pattern scan
Here is a strategic recommendation: [paste].
Check it against seven patterns and, for each, say
"present," "absent," or "cannot tell," with the
sentence that triggered your judgment:
1. Buried assumption: a conclusion whose required
conditions are not stated
2. Chosen baseline: a comparison whose reference point
flatters the case
3. Hockey stick: a projection that bends away from
history without a named mechanism
4. Friendly pilot: evidence from favorable conditions
generalized to unfavorable ones
5. Missing alternative: no do-nothing or staged case
6. Convenient source: evidence from an interested or
stale source
7. All-or-nothing ask: no gates, no stopping conditionsThe pre-mortem
It is [18 months] from now and this initiative failed.
Write the 3 most plausible failure stories, each with:
1. The specific chain of causes
2. The earliest observable evidence that this failure
was beginning (the tripwire), with a number and a
date where possible
3. What the reasonable response would have been at
that tripwire moment
Prioritize failure modes that stem from the
assumptions in my attack list.The murder board
Run a live hostile Q&A on this recommendation: [paste
one-pager or rebuilt document].
Rotate through four personas: a verifier who checks
numbers, a capital allocator focused on opportunity
cost, an operator attacking execution, and a historian
citing past initiatives: [abstracted history].
Ask one question at a time and wait for my answer.
After each answer, follow up once the way a skeptical
stakeholder would: probe the weakest part of what I
said. Every 5 questions, pause and score my answers
against this standard: direct answer first, evidence
second, implication third. Do not be polite. Begin.Standing instructions
The three courses' standing-instruction blocks, merged into one set. Add this to your AI tool's custom instructions, or to preferences.md if you built the second brain in course one.
When working with me:
Context and voice
- Read my identity and preferences files before any
substantive task
- Match the communication style described in my
identity file; default to shorter
- If I have not stated a position, ask me for one
before drafting; do not write around the absence
Accuracy
- Never invent facts, figures, names, or quotes; mark
gaps as [NEEDED: description]
- Before finalizing a draft, list every specific claim
you did not take directly from my input
- When a new decision contradicts an entry in my
decision log, flag it
Judgment and challenge
- Act as a red team by default on strategic
recommendations; do not validate
- Surface unstated assumptions before commenting on
stated ones
- Always ask "compared to what?" about a favorable
number, and ask for the reference class behind any
projection
- When I ask for a defense of my position, first give
me the strongest attack on it
- When I present a full commitment, ask what the
staged version would look like
Format
- No filler openers, no importance words without
content, no conclusions that conclude nothing
- Flag any sentence that exists for tone rather than
informationGlossary
- All-or-nothing ask
- A recommendation with no staged decision points and no defined stopping condition.
- Buried assumption
- A conclusion that depends on something never stated in the document.
- Capture test
- Capture only if you will act on it, you will reuse it, or it changes how AI should help you.
- Chosen baseline
- A comparison whose reference point was picked because it flatters the number, not because it's the honest one.
- Context gap
- The distance between what an AI model can do and what it knows about your specific situation.
- Convenient source
- Evidence from a party with a stake in the conclusion, or a study too old to still be trusted.
- Decision log
- A running file of what was decided, when, and why, with a "revisit if" condition attached to each entry.
- Friendly pilot
- A test run under the conditions most likely to make it succeed, then generalized past where it was tested.
- Hockey stick
- A projection that bends upward with no named mechanism for the bend.
- Master-and-cache rule
- Your files are the master copy of your context; any AI tool's memory or standing instructions are a cache refreshed from them.
- Missing alternative
- A recommendation argued without a real do-nothing case or a smaller staged version.
- Murder board
- A live, out-loud, hostile rehearsal of the hardest questions, run before the real presentation.
- One-pager
- A single page carrying a recommendation, its load-bearing assumption, the alternatives considered, the risks, and what would change the recommender's mind.
- Pre-mortem
- Imagining an initiative has already failed, then explaining why, which surfaces more specific risks than asking what could go wrong.
- Reference class
- The set of similar past efforts a projection should be anchored to before adjusting for what is genuinely different this time.
- Second brain
- A small, organized set of plain-text files describing your work, knowledge, and preferences, written so both you and an AI can use them.
- Signal Test
- Stakes, Specificity, Stance, Signature. Score any message 0 to 4.
- Stress test
- The five-step method, strip, rate, attack, compare, break, for pressure-testing a recommendation before anyone else does.
- Tripwire
- The earliest observable evidence that a specific failure story is beginning, paired with a pre-agreed response.
- Workslop
- AI-generated work that looks polished but lacks the substance to move a task forward.