What we're learning about maintained understanding.
Propperly is a bet that serious work should compound — that people and AI agents should build on what a project already knows instead of rebuilding it. We write to test that bet in public. Everything here is meant to be evidence-led: claims are scoped to what we can show, and articles stay in draft until they are.
Latest
The Picture Keeps Moving
The hard problem isn't giving AI agents more context — it's that organizations don't stand still while you capture them. A snapshot is not a maintained state of understanding.
Analytics practiceCan AI Replace Dashboards? What Changes—and What Doesn't
The useful question is no longer whether AI will kill dashboards. It is which analytics jobs are better served by conversation, which still need a persistent visual surface, and what has to be true before anyone should trust the answer.
- Why AI Analytics Still Gives Different Answers to the Same Business Question
Enterprise AI can have access to the warehouse, catalog, semantic layer and every relevant document—and still answer the wrong question. The missing piece is often not more retrieval, but a maintained view of what the organization currently means.
- Anthropic Got Its Analytics Agent to ~95% Accuracy. The Hard Part Wasn't SQL.
Anthropic reports ~95% analytics accuracy with Claude—and a fall to ~65% after a month without active Skill maintenance. The revealing part is everything they had to build around the model.
- How to Keep Claude Code Context Between Sessions
Claude Code already gives you several ways to carry context across sessions. The hard part is knowing which mechanism to use — and when remembering more stops being enough.
- Why the Same KPI Shows Different Numbers Across Dashboards—and How to Fix It
One dashboard says $4.8M, another says $5.1M, and both are labeled Revenue. Often nothing is broken — the dashboards answer different questions. A practical way to diagnose the gap and stop it from returning.
- What Is an AI Company Brain — and Why Should It Start in Analytics?
A company brain is not a chatbot over your documents. It is a maintained, human-approved current state of what your team knows — and analytics is the sharpest place to build the first one.
- AI Memory Is Not a Maintained Project State
"AI memory" and a maintained project state look similar and do opposite jobs. One recalls what was said; the other tells you what is currently true.
- Why Analytics Agents Need Decisions and Definitions — Not Just RAG
Retrieval finds text that mentions a metric. It cannot tell an agent which definition is current, which was ruled out, or whether the number is safe to state.
- Why Long-Running AI Projects Lose Their Own History
After a few months, the truth about a project lives in fragments — and neither search nor a summary reliably tells you what's still true.
- The Rejected-Paths Problem
The approaches a project deliberately ruled out are the first thing it forgets — and the most expensive thing to relitigate.
- Your CLAUDE.md Is a Maintenance Job
The context file you hand-write for your AI starts as a convenience and quietly becomes a second project you have to keep true.