Where instibrain sits.
Your pages are read whole, beside the index. What the team asks for again moves from the index to a page, and the brain grows from the questions.
new words
- Read whole
- Your AI reads a page from its first line to its last, in its context, before it works. No cut, no rank, no top five.
- The loop
- An answer found in the files, asked for twice, becomes a page where it belongs. instibrain basics, lesson 6.
the lesson
| A page | A chunk | |
|---|---|---|
| Reviewed by | The people of its space | Nobody; cut by a job |
| Read | Whole, every session | Five at a time, when a question matches |
| Cited by | Its path, in a space with a name | A file, a page number, a date |
| Fresh | On save | At the next run of the job |
| Who may read it | The space's readers, by role | Stored with the chunk, filtered before search |
| How many | Hundreds | Hundreds of thousands |
Neither replaces the other. A page cannot hold three years of tickets; an index cannot hold a position the team agreed on, because an index holds what was written, not what is true. The three levers put them side by side: retrieval is the second lever, context the third, and instibrain is the third.
The loop between them is the point. A question goes to the index because no page answers it. The answer is checked, and the second time the question comes, it goes on a page, with the file it came from as its source. From then on every AI of the team reads it whole, and the index is asked one question fewer.
try it
Type this and press Enter:
Which question did you have to search files or the web for today, instead of reading a page? Write its answer as a page where it belongs, with the file it came from as its source, and show me before you save.you should see
- A short page, in the folder whose purpose fits, its source named, and a question for your yes.
- If nothing was searched today: Claude says so. The loop runs on the questions pages cannot answer.