A principal picks up a phone before the day starts and has one question: what does today require of me? The answer fits in a document of twenty lines or so. Everything about the design of that document follows from a constraint that is not technical at all. The reader has one morning, one attention span and no intention of opening a mailbox to find out what the brief was about. A brief that is read once, at a fixed time, by a person with something else to be doing, is not a summary. It is a decision, taken in advance, about what will be read and what will not.
The distinction the whole structure turns on is the ordering. A feed orders by arrival, and it is honest about it: newest first. A brief that inherits arrival order is a feed with headings. This is not because the messages are unimportant; it is because the hour at which something arrived has no relationship to what the day requires. The invoice that arrived on Friday and is due on Wednesday, the contract that has been waiting for a signature since Monday, the meeting at eleven that requires a document nobody has read: none of these is more or less urgent for having landed at 07:12 rather than 19:40.
So the brief is best understood as a reading decision that somebody has already made. These things will be read and those will not. This will be read at 06:40 and that will be read when it is late. Made deliberately, that decision is a structure, and it can be argued with, tested and changed. Made by default, it is an inbox with a better typeface, and the reader is still doing the triage at the kitchen table.
This piece sets out the structure and defends it: the four actions the reader must be able to take, the five tests a line passes to earn its place, the five blocks in reading order with the cap on each, and a worked day in which fifty-three inputs become eighteen lines. It also says what a brief cannot know, which is the part most often left out of the design, and it is the part that decides whether the document is a map of the day or a map of the mailbox.
The test is what the reader can do
A brief is not judged by what it contains. It is judged by what the reader can do after reading it, with nothing else open. Four actions, and a brief that supports fewer than four has left work in the inbox rather than taken it out.
Say what cannot move. The appointment, the call, the deadline, the travel that makes any of them possible. If the reader has to open a calendar to discover at what time they must leave the building, the fixed points block has failed at the one thing it exists to do.
Settle what only they can settle. The two or three decisions where the answer is not in a document and not in anyone else’s gift: the price to quote, the candidate to advance, the clause to accept. A decision in this block is one the reader can take where they sit, because both options are stated and the consequence of each is on the line.
Answer or approve what arrived overnight. The reply that must go out in their name, the document that needs a signature, the name that has to be confirmed before somebody else can proceed. The brief does not write the reply; it puts the decision to approve or rewrite in front of the person entitled to make it.
Chase what is late. The thing that was due last Thursday and is still open. This is the block most likely to be skipped, because it is the block with no new information in it, and it is the block most likely to be the reason a brief exists. A late item does not become less late by being absent from the brief.
That is the reading test, and its virtue is that it can be failed in public. Give the principal the brief and nothing else and ask which of the four they could do. There is a second-order version that catches the commonest failure of all: after reading the brief, the reader must not need to open the inbox to find out what the brief is about. A brief whose lines each require a message opened behind them is an index. An index is a useful object and it is not this one.
What earns a line
Five tests, and a line that passes none of them is not short of the mark; it is not a line. They are worth applying in order, because the first is a veto and the rest are ties broken by the reader’s morning.
Consequence. Does the day change if the reader knows this? “Your flight is confirmed” changes nothing. “The flight is confirmed for Thursday, not Wednesday” changes Thursday. Consequence is not the same as seriousness: a serious subject that costs the reader nothing today is a matter for the file.
A date inside the week. Not “urgent”: a date. A line carrying a date can be acted on before the date becomes a problem. A line carrying “as soon as possible” has usually already been deferred once, and deferring it again costs nothing, which is exactly the difficulty.
An owner who is not the reader. A supplier who owes a delivery, a solicitor who owes a reply, a colleague who owes a signature. These belong in the brief precisely because the reader is the only person who can chase them without being asked to. A line naming someone else is a line the brief has done work on.
Irreversibility. Something that, once done, cannot be undone today: a signature, a payment, a public statement, a commitment in writing. Irreversibility is what makes a line worth reading at 06:40 rather than at eleven. Anything that can be corrected tomorrow is a line that can wait for tomorrow’s brief.
Novelty against the record. Not new to the reader’s memory — new to the record: the previous brief, the list tables, the calendar. A fact that has appeared in three consecutive briefs unchanged is either finished or it is a chase, and if it is a chase it earns its line by carrying the original date, not by being repeated.
What gets cut is a separate list, and the cut is a judgement about function rather than about importance. Everything that requires no decision and no action this week is filed and remains retrievable; it is not printed. Three emails about one meeting become one line, because the brief reports where the thread stands and not how it got there. Status with no change — “still waiting”, with no date — is cut, because it consumes a line to say nothing. Everything the reader wrote themselves is cut: they know, and a brief that reports their own sent items to them is charging for a mirror. Marketing, newsletters, notifications and calendar acceptances are cut as a class, and the class is not the brief’s to hold: it belongs in the triage table, where it is counted and capped and where somebody can see how much of the day it is. And anything whose content is a document is named rather than summarised: the brief says which document, what must be decided about it and by when, and it does not compress twelve pages of a contract into a line, because the compression is where the wrong sentence lives.
The proportion matters more than any single rule. On the worked day below, fifty-three inputs produce eighteen lines and thirty-five produce nothing. A brief of forty lines for the same day is not a fuller brief. It is a brief that has handed the reading decision back to the reader, who now has to do it at the kitchen table instead of over an evening.
The five blocks, in reading order
The order follows the reader’s morning, and each block answers one question. The cap is the number of lines the block may hold on a normal day; it is not a target.
Table 1. The five blocks of the daily brief in reading order, the question each answers, what a line in it must carry, the normal-day cap, and the material that must never appear in it. Source: constructed for this piece; the caps are a design decision, not a product specification.
| # | Block | The question it answers | What a line must carry | Cap | What must never appear |
|---|---|---|---|---|---|
| 1 | Fixed points | What cannot move today? | Time, place, who is there, and the travel or preparation that makes it possible | 6 | An entry that could be moved without consequence |
| 2 | Open decisions | What is waiting on this person? | The decision, the two options, the consequence of each, the date it must be taken by | 5 | A recommendation with no options behind it |
| 3 | Overnight | What arrived that needs a person this morning? | What came in, what it asks for, what it blocks, who is waiting on the answer | 6 | A message copied into the brief instead of a decision about it |
| 4 | Overdue | What is late and still open? | What was due, on what date, who owes it, and what happens if it slips again | 5 | An item that has been late for a month with no decision to drop it |
| 5 | Waiting on others | Who owes what, and when? | A name, what they owe, the date they promised | 4 | A promise with no date attached |
| Total | 26 |
Fixed points come first because a day is structured by what cannot move: the site visit at 09:30 determines when the reader leaves at 08:00, and both of those determine whether the 11:00 call can be prepared for. Open decisions come second because they are the only part of the day that will not resolve itself: everything else either happens or is chased, and a decision nobody takes stays open indefinitely. Overnight material comes third, ordered by whether it needs a person this morning, because it is the block most likely to expand without limit and therefore the block that most needs a cap. Overdue comes last, and the position is deliberate: it is the block the reader is most likely to postpone again, and putting it at the end of a capped document makes it the last thing read rather than the first thing dropped.
The cap total of twenty-six is not derived from anything. It is a promise, and that is the point of stating it: a structure that can grow without limit is not a structure, and a brief that occasionally runs to sixty lines has taught its reader that the first twenty are skippable. A cap that is breached is a signal — that the week is unusual, or that the triage rules upstream are wrong — and it is only visible as a signal because the normal number is written down.
The line is the unit
A line of a brief carries three things and often a fourth. What happened, in the record’s words rather than a paraphrase of them. What the reader must do about it. By when. And, where it applies, who is waiting.
Two lines about the same contract show the difference. “Contract from Halden for signature” is not a line. It has no date, no consequence and no decision in it; the reader’s only possible response is to go and find the email, at which point the brief has cost attention and returned nothing. “Halden contract: signature needed before Thursday’s visit or the deposit date moves to December; the changed clause is 7(b), the liability cap; options are sign as amended or return for a redraft by Wednesday” is a line. It can be acted on where it is read, and it can be argued with, which is a property worth more than it looks.
The structural consequence is worth stating plainly, because it is the part that a drafting model gets wrong when it is asked to be helpful. A line that cannot be written in that form usually means the underlying item has not been decided about yet, and that is information in its own right. It belongs in the brief as an open question — “the specification has been with the engineer since Monday and cannot be quoted until it returns; decide today whether to quote from the old specification or move the submission date” — rather than as a summary of a message that said the same thing in nine paragraphs. The brief’s job with an undecided item is to make the decision visible, not to make the item sound handled.
A worked day
The day below is constructed so that its arithmetic can be followed. It is a Tuesday, the day after a quiet Monday, and the principal is a small property developer with four live projects, one lender, one solicitor and a staff of six.
Table 2. One constructed working day: the inputs, the lines they produced, and where everything else went. Source: constructed for this piece; every count is stated so that the total can be checked.
| Input | Count | Lines in the brief | Where the rest went |
|---|---|---|---|
| Messages in the overnight tray | 38 | 9 | 3 of the 4 immediate cases to the overnight block (the fourth was telephoned before the brief was written); 4 of 7 decisions to the decisions block; 2 of 6 waiting items to the waiting block; 11 answered rows, 2 routed to a named person, 5 read, 3 filed in the triage table |
| Calendar entries | 6 | 5 | One travel block merged into the visit it serves, which is one line and not two |
| Open items on the list tables | 9 | 4 | 5 not yet late: due later in the week and not today’s problem |
| Total | 53 | 18 | 35 inputs produce no line |
The eighteen lines were these. Five fixed points: travel at 08:00, the site visit at 09:30 with the surveyor, the lender’s call at 11:00, second interviews at 14:00, board papers due at 16:30. Four decisions: the Halden signature, the quote at 12 per cent against 8 per cent with a longer warranty, the third interview slot, and whether to renew the equipment lease for two years or three. Three overnight items: a tenant’s complaint about the lift, a supplier’s delay on the switchgear, the accountant’s request for a document that has been asked for twice. Four overdue items: two invoices past thirty days, one signature that has been outstanding for nine days, one specification that has not come back from the engineer. Two waiting items: the solicitor’s reply on title, promised for Monday, and the surveyor’s report, promised for today.
Eighteen lines, twenty-six permitted, and the arithmetic of the cut is the interesting part. Of the thirty-eight messages, twenty-nine produced no line at all. Eleven were answerable from a document and became rows with a suggested reply — the reader will approve sentences, not compose them. Five were newsletters and notifications and were read in a count rather than in a line. Three were filed as records. Six named somebody who owes something, and only two of those carried a date close enough to matter today. Four arrived needing a person immediately, and three of them are in the brief because they will still need a person at nine.
The cost of this day is two paid streams: the brief itself, at two calls, and the triage of thirty-eight messages, at one call a batch of ten, which is four batches. The brief is $0.01400 and the triage $0.02800, so the day is six paid calls and $0.04200. Per line of brief, at eighteen lines, that is $0.00078; per message triaged, $0.00074. Over twenty working days the role’s month is 170 paid calls and $1.19000, of which the brief is 40 calls and $0.28000 — 23.5 per cent of both the role’s calls and the role’s cost, because a single rate makes the two shares equal. A year of briefs at the same pace is 480 calls and $3.36; the whole role for a year is 2,040 calls and $14.28.
One structural property of the arithmetic is worth keeping. A batch is ten messages, so thirty-eight messages and forty messages cost the same four calls, and the thirty-ninth message of the day is free while the forty-first is not. At the margin the instrument is nearly free and then suddenly is not, which is a reason to set the batch size deliberately rather than to discover it by accident on a heavy Monday.
What the brief cannot know
Four limits belong to the design rather than to any implementation of it, and the first is the one a principal discovers for themselves within a fortnight.
It knows only what it is fed. The inputs are the mailbox, the calendar and the documents sent to it. A decision taken in a corridor, a call that changed the plan, a commitment made at dinner: invisible until somebody writes it down. The brief can be entirely honest about this and still mislead, because the failure is in the reading. A brief is a map of what was written down, and a principal who treats it as the day will be surprised on the day the plan changed in a doorway.
A cap is a decision about what will be read, and a cap can bury. An item with a date three weeks out is nobody’s line today, and it is still nobody’s line in a fortnight unless something makes it so. This is why the test for a line is a date and not a feeling of urgency: a dated item can be promoted by rule as its date approaches, and an urgent item cannot, because urgency has no schedule. A structure that sorts by consequence and then forgets is a structure with a horizon of about five days.
The brief reports; it does not decide. A decisions block that arrives with recommendations has moved the decision to the instrument and left the principal reading a decision rather than making one. The line states the options and the consequence of each, and states no preference. This is a limitation of the artefact, not a modesty: the principal has context the brief does not, which is the whole reason the brief exists.
Length is not honesty. A short brief after a quiet night is a good brief. A short brief after a heavy night is a blind spot, and the document itself cannot tell the two apart. The only instrument that does is the triage table read against the inbox, once, by a person — the same audit that a support desk runs on its queue.
The brief is not correspondence
The brief is an internal document. It is written for one reader, that reader is the principal, and it leaves the workspace as a file or as a message to that reader’s own paired device. Nothing in it is addressed to a third party, and nothing in it is drafted in anybody else’s name.
The suggested replies that sit beside it are a different object. Each is a draft in an outbox, carrying the reason it is in the outbox, and each waits for an approval. Approving a sentence takes seconds where composing one takes minutes, and that difference is real; it is not a licence to send. Sending is a separate permission from drafting, the mailbox is under an allow-list, and no instruction inside a message — including one that says to just send it — converts a draft into a send. A brief with an automatic send beside it is not a stronger brief. It is a different instrument with a different risk, and the risk lands on the person whose name is on the reply.
The decision
Three gates, and they are cheap enough to run weekly.
The reading test. Give the principal the brief and nothing else. Can they state what cannot move, settle what is open, answer what arrived and chase what is late? If one of the four fails, name the block that failed, fix that block this week, and leave the rest alone.
The line test. Read every line and find what the reader must do in it, and by when. A line without both is a headline, and it is cut or rewritten into a line. This is the only editorial rule the document needs, and it is the rule that the cap enforces, because a brief that has to stay under twenty-six lines cannot afford many headlines.
The audit. Once a week, read the triage table against the inbox and answer one question: did anything of consequence arrive and not reach the brief? A quiet brief after a heavy night is either a good night or a blind spot, and this is the only test that distinguishes them. When the answer is a blind spot, it is almost never the drafting that failed. It is a rule upstream that put the item in the read class, or a cap that kept it out of the sample, and the fix belongs to the rule set rather than to the brief.
Where this goes
- The pillar: x402.review/2026/10/31/front-office/ — the family, the two roles and the boundary this piece works inside
- The role: studio.x402.press/roles/personal_assistant — the brief, the triage table and the send that waits for approval
- The service it buys: x402.press/services/deepseek/ — the canonical price, seven chains
- The service it buys: x402.press/services/qwen/ — the second model, priced per call
- The reviews: x402.review/2026/10/31/deepseek-review/ and x402.review/2026/10/31/qwen-review/ — the measured handshakes and what they do not promise
The cost of a brief’s month
The brief is two calls a working day at $0.00700, the rate the DeepSeek endpoint asks on Base under exact: $0.01400 a day, $0.28000 over twenty working days, $3.36 over a year of briefs. The role’s whole month is $1.19000 across 170 paid calls, and the brief is 23.5 per cent of it.
Fifty calls at the measured rate is $0.35000, which is twenty-five days of two-call briefs. The figure that circulates as DeepSeek’s price, $0.00875, is Qwen’s measured rate and not DeepSeek’s: at it the same fifty calls are $0.43750, $0.08750 more, and the brief’s month would be $0.35 rather than $0.28. At the metered ceiling of $0.00955 — a long prompt, or upto at a thousand authorised output tokens — the forty brief calls of a month are $0.38200 rather than $0.28000. The ceiling is not a price rise; it is the difference between asking for a short document and authorising a long one.
Three lines of the role’s month are not in this figure and are not metered: the document parts — the engine that renders the brief, the reading floor that opens what arrives, the data table that holds the triage rows — and the two connectors that carry the mailbox and the delivery. None of the three carries a published per-call price, and there is no 402 challenge to read for any of them. Their absence understates the month rather than discounting it. Where a supplier publishes no per-call price, the correct entry is not an estimate; it is the statement that no price exists.
Method note
Prices were read from the live 402 handshake on Base on 31 October 2026: $0.00700 for a short prompt under exact, $0.00955 for a long prompt under exact and as the upto ceiling, and $0.00875 for Qwen’s qwen3.8-flash under both exact and upto. The month’s call counts — two calls a brief, one call a batch of ten messages, one call a day for the itinerary and list tables, three calls a meeting — are the role page’s planning assumptions, not observations, and the day in Table 2 is constructed to the same assumptions so that the arithmetic can be followed.
Table 2 reconciles: 38 messages plus 6 calendar entries plus 9 open items is 53, and 9 plus 5 plus 4 is the 18 lines, so 35 inputs produce no line. The nine lines from messages are 3 immediate cases plus 4 decisions plus 2 waiting items, and the remaining 29 messages are 1 case telephoned before the brief, 3 decisions held for the triage table, 4 waiting items without a date inside the week, 11 answered rows, 2 routed, 5 read and 3 filed, which is 29. The cost check: 2 plus 4 calls is 6, and 6 times $0.00700 is $0.04200; 18 lines at $0.01400 is $0.00078 a line; 40 calls a month is $0.28000 and 170 calls is $1.19000, and 0.28 divided by 1.19 is 23.5 per cent, as is 40 divided by 170.
No time figure appears in this piece as a quantified claim. There is no minutes-saved figure, no response time and no hours-recovered estimate anywhere in it, because none exists: the only quantities measured here are calls and their prices, and the caps in Table 1 are a design choice rather than a measurement of anything. The single comparison of effort the piece makes — approving a sentence against composing one — is qualitative, and it is the pillar’s formulation rather than a measurement. No client, principal, tenant, supplier or message is described, and none is real.
x402.review is written and edited by N.P. Vincent. The publication is connected to x402.press, which sells the services reviewed here; that interest is disclosed on every piece that touches a service, and it is stated here rather than left to inference.
Filed under Jobs · Tags: daily-brief, personal-assistant, structure, prioritisation
