Skip to content

Blog

Articles about AI, research, product management, and working with technology.

How many names can one respondent evaluate: a methodology breakdown

There is no universal answer — the threshold depends on format. Three name-testing methodologies with numbers from Sawtooth, Ipsos, Qualtrics, and SurveyMonkey.

research-methodsname-testingconcept-testingmaxdiff

How AI learns to reason: the race to teach models where they went wrong

Six teams are racing to solve one problem: AI models cannot tell which step in their reasoning mattered. Here is what they found and why it affects every prompt you write.

ai-reasoningmachine-learningreasoning-models

Four schools of Jobs to be Done: how a milkshake story became four research methods

The phrase Jobs to be Done hides four distinct approaches to understanding customers. A guide to choosing the right one — with a mini periodic table.

research-methodsjtbdproduct-research

Axios supply chain attack: why the advice going around only checks macOS

The popular check-yourself commands for the axios npm compromise only cover macOS. Here is the full technical breakdown and verification steps for all three platforms.

securitynpmsupply-chain

Claude Code source has been available for 13 months, and nothing happened — why?

The full source code of Claude Code has been in the npm package since February 2025. Researchers extracted the roadmap, prompts, and hidden features — yet the product keeps growing.

ai-toolssecurityopen-source