The Loop

The loop

brainless is one loop, and each part is there because the one before it is useless alone. Notes that nobody reads are a warehouse. Summaries that nobody argues with are a tidier warehouse. And an argument that nobody grades is entertainment.

The README tells it as four lines. This page walks the same four lines with the mechanics and the folder names.

flowchart LR
  G["1. Gather information<br/>capture everything, the machine reads"]
  T["2. Think deeper and clearer<br/>ask what you know, six voices argue"]
  D["3. Decide<br/>write a bet, grade it later"]
  X["4. Bonus: get shit done<br/>promises become tasks, three a morning"]
  G --> T --> D --> X
  D -->|"filed back, so tomorrow builds on today"| G

Loading

The bet underneath all four is connectivity. A note on its own is storage. Notes wired to each other are a map, and the value shows up when you move across it: an argument from a meeting in May lands beside a belief written in March, and the two change each other.

1. Gather information

Capture everything, filter later

Handwritten pages, voice notes, photos, links, documents, half sentences, Telegram messages and Buzz #inbox posts all come in through the same wide front door. They land in Inbox/, Inbox/Links/ and Thinking/Daily/ as Markdown, and nothing is judged on the way in. Judging a thought at capture time is how good ones die in a car park. The Capture Flow page maps every entrance and every scheduled job.

Let the machine read so you do not have to

A pile of captures is only useful if somebody reads it. So a compiler turns every note into a summary, links it to the people, companies and ideas it mentions, folds it into concept pages that update in place (disagreements kept side by side, old claims struck through, never deleted), mirrors project status and writes a daily digest into .wiki/. YouTube and podcast links come back as timestamped transcripts first. Documents keep their originals and their raw Markdown conversions. The high-value ones also get a Summary and a Fiche de Lecture, and a missing or empty output is retried instead of quietly accepted.

Obsidian is the surface. Every input becomes Markdown, Graph view follows the explicit [[wikilinks]], and Smart Connections shows related notes that do not have a written edge yet. You own the notes. The machine owns .wiki/ and can rebuild it from scratch whenever it likes, so nobody has to spend a Sunday curating a knowledge graph. Why everything becomes Markdown, and the Zettelkasten underneath: Method.

2. Think deeper and clearer

Ask what you already know

Before anything hard, brainless search or a slash command answers "what have I already written about this", including the parts you had forgotten writing. The Commands page lists every one, with the thinking error it is built against.

Argue before you act

Six personas test a thesis in two rounds. In round one each persona gets its own thread, applies its own method without seeing the others, and votes YES, NO or CONDITIONAL with a probability. In round two each reads the rest, picks the strongest objection, answers it, votes again and names the new evidence that moved it, or says "none".

A script scores the round before the moderator writes a word, so the synthesis starts from counted votes and not from a mood. A rolling 30 day scorecard raises a flag when affirmation climbs above 60 per cent, or when a persona never votes NO. Six voices that always agree with you are a fan club.

Full detail, including the book behind each voice: Personas and Dialectic. Nothing here makes you smarter; it changes what you have to write down before you may move on, and the order you see things in. Each mechanism: Thinking Clearer.

3. Decide

Write what you believe and what you decided, as bets

Reading is still not thinking. The compiled layer can tell you what you wrote, but it cannot tell you what you hold to be true, so that part is typed by hand. Beliefs live in Thinking/Beliefs/, each with a line for "what would change my mind". Decisions live in Thinking/Decisions/, each with a prediction, a confidence and a review date.

The format matters more than it looks. A bet can lose, and something that can lose can be argued with.

Grade yourself

An argument ends in a bet, and a bet is worth nothing until somebody settles it. It never picks for you. When a review date passes, the system nags until you write the outcome. brainless calibrate lists what is due.

Ten years of ungraded decisions is one year repeated ten times. If you do only this step, you are ahead of most people. Every analysis, graded or not, is filed back into .wiki/digests/queries/. That is where the loop closes, because tomorrow's question gets to build on today's answer.

4. Bonus: get shit done

Decisions come with promises attached, and promises are the first thing to fall out of a busy head.

• Promises become tasks. Commitments from meeting reports and the day's captures are appended to the task ledger in _Agent-Context/TASKS.md, with a near-duplicate check so a promise rephrased on a later day does not become a second task. An optional sync keeps the ledger and Google Tasks in step both ways.
• At most three things a morning. brainless today offers a decision due for grading, the oldest open commitment and one piece of evidence to review, each with a link to its source. Empty categories stay empty, and finishing an item does not refill its slot. See Today Queue.
• Answer in a thread. The queue, reminders and the weekly question arrive in Buzz. Answers are previews until you apply them: "apply" writes the note, "defer to Friday" moves it, "dismiss" drops it with a reason. See Buzz Interactions.

---

Who this is for

People who make decisions for a living and suspect their own reasoning. Founders, operators, product people, anyone whose calendar is a sequence of judgement calls. It helps if you already write things down. It helps more if you are willing to be told you are wrong by a script at 21:20.

What it is not

• Not a chat app, although it is getting conversational. The notes are still the record. The threads are how the record gets made.
• Not a knowledge graph you maintain. The graph is built for you, not curated by you.
• Not a decision maker. It never picks. Duke would call that resulting in advance.
• Not finished. Built for one person, then made installable in a day. Expect rough edges.