HeyJaredResearch toolbox
Understanding people, measuring what matters, building for humans. UX Research, product thinking and AI evals.
- Indexed issues, last 90 days
- 17
- Latest publication
- Sep 30, 2026
- Audience
- Checking…
- Earliest in this view
- Aug 16, 2026
Latest issues
Your AI Agents Need an Org Chart (opens the original)
Read excerpt
RESEARCH . 41We still ask too much of a single agent: understand the problem, use the tools, do the analysis, check the work, resolve uncertainty, and produce the answer.For complex work, splitting that responsibility can be effective. One agent can search while another analyzes. A specialist can work in a clean context. Several agents can examine the same evidence separately. A reviewer can look for problems the original agent was not trying to find.But multi-agent systems start in the wrong pl
How to Build an AI Eval That Represents Real Product Use (opens the original)
Read excerpt
AI EVALS . 5In the earlier parts of this series, we worked through what we want an eval to measure, how to turn that into tasks, and how to grade the results. The next question is which tasks should actually go into the eval.Sampling and coverage are closely related, but they can serve very different goals. An eval designed to estimate performance under actual product use needs a different task set from an eval that is designed to deliberately detect failures. There are other sampling goals too.
The Conditions Under Which We Like Ourselves (opens the original)
Read excerpt
RESEARCH . 40Today’s essay: A short reflection on how we understand people and how the interactions and environments we share with them help shape the person we come to know.<div class="image2-i
What Synthetic Users Are Good For? (opens the original)
Read excerpt
MODELS OF PEOPLE · 03In the first two parts of model of people series, I argued that synthetic users are not one method and what we can learn from them depends on the goal, what is being represented, where that representation comes from, and what kind of system produces the behavior.The debate around synthetic users often gets framed too absolu
When Knowing How Is No Longer Required (opens the original)
Read excerpt
RESEARCH . 39I barely know anyone’s phone number anymore. I can follow GPS through a city and realize later that I could not reconstruct the route. I often remember that I read something useful without remembering the fact itself because I know roughly how to find it again.We usually notice these changes as losses. We remember fewer phone numbers, construct weaker mental maps, and retain less information that can be retrieved in seconds. But another explanation fits the same pattern: <stro
Publishing over time
Last 90 days. Choose a month to open its work.
Recurring subjects
Named in the text we hold. One piece can cover several.
Audience
No verified audience measurement yet.
About this data
Counts cover the work we have indexed. Tone needs enough text and a confident classification. Excerpts and episode notes are not full articles or transcripts.
Identity or attribution wrong? Suggest a correction.