~/articles
Studies and notes on AI, Machine Learning, architecture, MLOps, AgentOps, frameworks and methodologies.
What an AI harness is, the five pieces it puts around the model (context, tools, limits, security and evaluation) and the questions to design your own.
8 min read
Where basic RAG fails, which failure each of the five architectures fixes, what it costs and how to evaluate it before going to production.
31 min read
How to combine products and markets, existing or new, to map growth options, compare risks and test each path with a cheap experiment, with a prompt for LLMs.
6 min read
How to calculate relative market share, position offerings by market growth and discuss resource allocation without letting stars and dogs decide for you.
7 min read
How to map the factors an industry competes on and use eliminate, reduce, raise and create to design a different value proposition, with hypotheses to test and a prompt for LLMs.
6 min read
Period-by-period cash flow, NPV, ROI, TCO and sensitivity in a build vs. buy example: how to assess whether benefits justify costs and separate fact from estimate.
11 min read
How to build the activity network, run the forward and backward passes, find float and shorten the schedule at a known cost, and why resources and uncertainty change the CPM answer.
12 min read
Elimination criteria, scales, weights and sensitivity: how to build a weighted matrix that exposes trade-offs instead of hiding opinions in a spreadsheet.
11 min read
How to cross urgency and importance to decide what to do now, schedule, delegate or eliminate, with clear criteria, a tech lead example and a prompt for LLMs.
6 min read
How to break a problem down to what's factual, investigate each assumption and rebuild the solution, with a full data processing cost example and a prompt for LLMs.
11 min read
How to use the cause-and-effect diagram to organize candidate causes, flag what has evidence and decide what to investigate first, with a software adaptation and a prompt for LLMs.
6 min read
How to run the 5 Whys with evidence at every link, parallel chains and a focus on the system, and when the technique falls short, with a prompt for LLMs.
7 min read
How to map what drives and what holds back a change, choose where to act and reduce barriers instead of just adding pressure, with an AI adoption example and a prompt for LLMs.
7 min read
How to define a hypothesis, metric, guardrails and sample size, calculate an A/B test and read the result without confusing a p-value with certainty, in a full case of falling activation.
12 min read
How to use inversion, provocation, analogy and recombination to generate alternatives and assess viability only afterwards, with a logistics fuel-saving example and a prompt for LLMs.
8 min read
How to define the universe, criterion and boundary rules to break a problem into parts that don't repeat and leave nothing out, with a funnel example and a prompt for LLMs.
6 min read
How to use observe, orient, decide and act when the situation changes as you decide, why orientation is the center of the loop and a full incident response case.
11 min read
How to analyze entrants, suppliers, buyers, substitutes and rivalry through each force's mechanism, tie industry structure to margins and avoid confusing the model with SWOT.
14 min read
How to run a pre-mortem in a short session, turn reasons for failure into prioritized risks and preventive actions, and what the evidence really says about the technique.
8 min read
How to build scenarios from critical uncertainties, test each decision against different futures and set warning signals, without confusing a scenario with a forecast.
13 min read
How to use Edward de Bono's six hats in meetings: parallel thinking, sequences for each situation, the blue hat's role, a subscription switch example and what the evidence says.
8 min read
How to separate internal and external factors, prioritize with evidence and cross the quadrants to reach actions, with a prompt for using LLMs as support.
8 min read
From temporal feature engineering to survival modeling: how to compare architectures, avoid leakage and build reliable probabilities of default.
19 min read
How to structure complementary perspectives to support decisions, creativity and critical analysis with generative AI.
24 min read