← Home
I Have This Problem #
Find the right framework, tool, or vocabulary term for your AI deployment challenge. Each answer links to the full concept with worked examples and the math behind it.
[“I keep hearing "knowledge work is manufacturing" - but how do I tell which kind of factory my line actually is?”
The Factory Typology →Framework
archetype = f(input, process, output). Name your line's signature and inherit a mature physical industry's operations playbook.](/frameworks/factory-typology/)[“My team runs every workflow with the same QC. Why does that keep failing on some lines?”
What Factory Are You? →Tool
Six questions classify your line. Most lines are chains with a curvature sign-flip: screen upstream, control downstream. One posture across the whole thing is wrong.](/tools/factory-typology/)[“We keep shipping improvements and then regressing. How do I make sure quality only goes up?”
Quality Ratchet →Lexicon
A CI-enforced floor that only moves up. Each improvement becomes the new minimum.](/lexicon/#quality-ratchet)[“How do I make AI agents improve over time without writing an improvement plan?”
Quality Hillclimb →Framework
Ratcheted quality gates on stochastic output create emergent ascent. The agent doesn't need a plan.](/frameworks/quality-hillclimb/)[“We deployed AI but we're scared to remove the human. How do we safely give it more independence?”
The Promotion Protocol →Framework
A 3-state progression: Disabled, HITL, Autonomous. Promote on statistical evidence, roll back on drift.](/frameworks/promotion-protocol/)[“Nobody agrees what "good" looks like for this task. How do I define quality?”
The Performance Frontier →Framework
Map the distribution of human performance, find the 99th percentile, compute the gradient toward it.](/frameworks/performance-frontier/)[“My AI keeps doing things I don't want. I can't write a complete spec of what I want.”
The Deity Problem →Framework
Three evidence channels: structured elicitation (conjoint), revealed preference (behavioral observation), and direct query (ask when VOI exceeds attention cost).](/frameworks/deity-problem/)[“Should I automate this specific task? How do I decide?”
TaskVector →Tool
Score across 9 dimensions. If any single dimension is a landmine (1), it's a hard no.](/tools/taskvector/)[“Is this task a good candidate for AI? What is the ROI?”
Verification Quadrant + Ablaza Ratio →Tool
T = time_to_do / time_to_check. High T = AI creates leverage. Low T = you're doing the work twice.](/tools/verification-quadrant/)[“The AI output looks good but I don't know if it's correct. Checking takes as long as doing it.”
Verification Trap →Lexicon
Easy to generate, hard to verify. T approaches 1. You have added a step without saving time.](/lexicon/#verification-trap)[“How do I brainstorm business ideas that actually match real demand?”
The Demand Field →Framework
Fix demand (immutable), vary means (mutable). Demand is a hidden force on your optimization gradient.](/frameworks/demand-field/)[“How do I guarantee this initiative succeeds instead of hoping?”
Designed Convergence →Framework
Design the game so rational agents converge to your outcome. Finite state + Bayesian search + ratchet = structural guarantee.](/frameworks/designed-convergence/)[“What are the actual dollar costs of AI being wrong?”
Dollarized Confusion Matrix →Tool
Replace accuracy with dollars. Optimal threshold: theta* = C_FP / (C_FP + C_FN).](/tools/dollarized-confusion-matrix/)[“How do I tell if an AI roadmap idea is actually good before committing to it?”
The DoG Test →Tool
A 60-second Claude prompt. Three checks (strength, surprise, utility). Verdict: GOOD DoG, NEEDS TRAINING, or BAD DoG.](/tools/good-dog/)[“The task is in a bad quadrant. What capital investment moves it to a better one?”
Quadrant Shifting →Tool
Five moves: build a verifier, decompose, enrich inputs, constrain outputs, build a rubric.](/tools/quadrant-shifting/)[“I have 20 projects competing for budget. How do I decide which ones to fund?”
Capital Allocation →Framework
Treat each project as an investment instrument with a return distribution. Rank by Sharpe (not NPV), map risk tolerances, handle correlations, construct the efficient frontier.](/frameworks/capital-allocation/)[“I know this project should be killed but I can't get organizational buy-in to stop it. How do I say no?”
Kill Protocol →Framework
Three decision gates: marginal Sharpe (is it additive to the portfolio?), left-tail survivability (can you absorb the downside?), base rate verification (does the team's P(success) match reality?). Portfolio math gives you the language for "no."](/frameworks/kill-protocol/)[“I have a budget and a set of candidate projects. How do I know which combinations are efficient?”
The Operating Efficient Frontier →Framework
Markowitz (1952) applied to operating investments. The non-dominated set of portfolios at each level of risk. By definition, points below the frontier are dominated - same risk, lower return, or same return, higher risk.](/frameworks/efficient-frontier/)[“Two allocators with the same inputs pick different portfolios. How do I know which is right?”
The Risk Tolerance Map →Framework
Both can be right. The map translates firm-level constraints - cash, runway, covenants, narrative - into the preferred point on the frontier. Without it, two rational allocators diverge on identical inputs.](/frameworks/risk-tolerance/)[“We shipped four projects and they all slipped the same week. How do I plan so that doesn't happen again?”
Correlated Execution Risk →Framework
Shared team, stack, review queue, or narrative means shared failure mode. Operating portfolios aren't sums of independent projects. The covariance matrix is the input that makes Stage 3 different from project-by-project NPV.](/frameworks/correlation/)[“Where is value leaking in my business that nobody has named?”
Directed Graph + Soft Spots →Framework
Your chart of accounts is a directed graph. Walk the edges. The soft spots are where value leaks.](/frameworks/directed-graph/)[“I spend all my time in meetings and firefighting. How do I invest in systems?”
Compile Time vs. Runtime →Lexicon
Compile time: building systems with multiplicative ROI. Runtime: executing tasks with single-period returns.](/lexicon/#compile-time)[“Is this AI automation a wasting asset or a compounder?”
Dual Curve + Knowledge Capital →Lexicon
Models depreciate while data appreciates - the net rate determines the investment type. See also: knowledge-capital framework.](/lexicon/#dual-curve)[“What is the NPV of automating this task? Should I build, buy, or hire?”
Automation NPV →Tool
Calculate NPV, IRR, and payback period against hiring, SaaS, or doing nothing. Same math, different asset class.](/tools/automation-npv/)[“How should I think about AI output I can't observe directly? How does the agent learn what I want?”
Structured Elicitation, Revealed Preference, Direct Query, Drift Detector →Lexicon
The four evidence channels from The Deity Problem: structured-elicitation (conjoint), revealed-preference (behavioral observation), direct-query (ask when worth it), and drift-detector (posterior predictive check). See also: oracle-gradient, designers-seat.](/lexicon/#structured-elicitation)[“How do I infer what the operator actually wants without asking? How do I learn from behavior instead of instructions?”
Revealed Preference →Lexicon
Watch what the operator does, not what they say. Revealed preference theory (Afriat, GARP) applied to AI alignment. Cheapest evidence channel because the operator behaves naturally.](/lexicon/#revealed-preference)[“When should the AI ask the operator a question? How do I avoid over-asking or under-asking?”
Direct Query →Lexicon
Ask only when the expected value of the answer exceeds the cost of the operator's attention. An agent that asks too many questions is poorly calibrated, not diligent.](/lexicon/#direct-query)[“How do I detect when the operator's preferences have changed and the model is stale?”
Drift Detector →Lexicon
A posterior predictive check on recent decisions. When the fraction the model predicted incorrectly exceeds a threshold, trigger re-elicitation. Preferences drift - the system must detect it.](/lexicon/#drift-detector)[“What does "operational alpha" mean? How is it different from just doing a good job?”
Operational Alpha →Lexicon
Excess return on enterprise value generated through systematic identification of mispriced edges. The directed graph finds them. The tools evaluate them.](/lexicon/#operational-alpha)[“What is the AI Sweet Spot? When does AI create the most value?”
AI Sweet Spot + Ablaza Ratio →Lexicon
Hard to do, easy to check - T >> 1. The templeton-ratio measures the gap, and a high ratio means transformative ROI. The proof-layer and gold-standard define what "correct" means.](/lexicon/#ai-sweet-spot)[“What should I build first - the AI system or the verification instrument?”
Proof Layer + Gold Standard →Lexicon
Build the rubric first. The gold-standard IS the verification instrument - without it, you're measuring with a broken ruler. See: soft-spot for finding where to look.](/lexicon/#proof-layer)[“How do I manage the pull of real demand on my product trajectory?”
Demand Gravity →Lexicon
The inescapable pull of real demand on product trajectories. Map it or crash into it.](/lexicon/#demand-gravity)[“Should I be playing the game or designing it? What is the CTO's real job?”
The Designer's Seat →Lexicon
Design the game so self-interested agents produce the outcome you want. Most engineering is game-playing. Mechanism design is game-designing.](/lexicon/#designers-seat)[“The AI deployment is at autonomy-state-machine HITL state. When do I promote to runtime autonomous?”
Promotion Protocol + Autonomy State Machine →Lexicon
Consecutive batches below acceptance threshold. Not one good batch - N consecutive. The construction-spread is the gap between build cost and operational value.](/lexicon/#autonomy-state-machine)[“How do I measure what I should invest in next for my AI operations portfolio?”
Knowledge Capital + Permutations →Framework
Knowledge work either compounds or depreciates. Invest in the appreciating side: verifiers, data, rubrics. Not the depreciating side: models.](/frameworks/knowledge-capital/)[“Why is a verifier worth building when it costs more than the generator it wraps?”
The Capital Value of Verifiers →Framework
A verifier is one of the only capital assets that appreciates through operating use. Every failure caught is encoded; the next run inherits the catch. Most operating assets depreciate; this one ratchets up.](/frameworks/verifier-capital/)
The Three Layers #
Frameworks tell you where to look - strategic models for finding enterprise value.
Tools tell you how to evaluate - interactive calculators for specific decisions.
Lexicon gives you the language - vocabulary that travels in meetings you are not in.
See also: Frameworks · Tools · Lexicon