Something arrives
A market move, a company filing or a project actual lands from a scheduled feed, with its source and date.
- Home
- Product
- The brain
The brain · continuous monitoring
Watches the market. Waits for a person.
The brain runs on a schedule across every workspace. When a market move, a filing or an actual arrives, it finds the exposed projects and holdings, prices the effect in money, drafts a proposal and verifies it in a sandbox. Then it puts the proposal in the right person's queue, with its evidence. Nothing changes the plan until that person approves it.
Status: partial. The scheduled pipeline from signal to approval queue runs today. Learning from outcomes needs months of real decisions, so that part is still ahead.
One fixed sequence
From a signal to a decision someone owns
Every scan follows the same steps in the same order, so a proposal always arrives with the same things attached: what moved, who is exposed, how much money is at stake, what the engine found when it checked, and how confident the brain is.
01 · Signal
02 · Expose
Who is exposed
Capibud identifies every project and holding that loads on that factor, across all four decisions.
03 · Price
What it costs
The effect is priced in money, at the P50 and in the adverse case, not as a red, amber or green light.
04 · Draft
What to do
A proposal is drafted: re-phase, hedge, add contingency, slow the commitment pace or reopen a decision.
05 · Verify
Does it hold
The engine re-runs the proposal in a sandbox. If its result contradicts the proposal's claim, the proposal is flagged rather than forwarded.
06 · Route
Who decides
The proposal reaches the right person's queue through your delegation of authority, with its evidence and a calibrated confidence.
07 · Approve
A person signs
Nothing changes the plan until a person approves it. The approved change is previewed, applied and can be undone.
08 · Record
It is remembered
The proposal, the decision and the reason are written to the append-only record, and the reason feeds the brain's calibration.
Watch it work
Watch a signal move
through the book
An illustrative copper move travels through one cable supplier to three projects, is priced, drafted, verified and routed, and then waits for a person. Double-click any node to see what is at risk and why.
Illustrative example. The figures in the player are not taken from a client project.
What it watches
The outside world, mapped to your book
A price move matters only if money in your book loads on it. The brain keeps a map of which projects and holdings are exposed to which shared factor, so a signal lands on the right lines and nowhere else.
Market prices
Commodity, rate and currency series on a schedule. A versioned rule fires when, for example, a commodity moves 1.5 standard deviations.
Company filings
SEC 10-K, 10-Q and 8-K filings and XBRL company facts, fetched on a schedule, compared with the previous filing and passed to the brain when a figure changes.
Project actuals
Cost ledgers and monthly reviews update the P80 cost and finish of each unit, so drift shows up as money at stake rather than a late surprise.
Contractors and counterparties
A contractor's listed peer falling 25% is a known warning pattern. Counterparty health and supplier concentration are scored, so one supplier carrying three projects is visible.
Committee tripwires
The thresholds a committee set when it decided are checked on every scan. A breach, or a close approach, becomes a proposal to revisit the decision.
Capacity and calendars
Tracked competitor capacity is compared with the market balance the engine assumes, and weather and logistics windows are checked against the work still to do.
How it reasons
Three kinds of model, each with a narrow job
Rules catch what you already know to watch for. Fitted models estimate whether a move will matter. Language models, if you switch them on, read text and explain. None of them is allowed to produce the numbers on its own.
Versioned rulesFast, explicit
Thresholds that catch known patterns quickly, such as a commodity moving 1.5 standard deviations or a contractor's listed peer falling 25%. Each rule carries an identifier and a version, so every proposal can say which rule fired.
Has something we already know to watch for just happened?
Fitted modelsFitted, then frozen
Reference-class overrun priors, signal-to-probability calibration and factor correlations, fitted and then frozen. The signal-persistence model has an out-of-sample Brier score of 0.172, where a coin flip scores 0.25.
Will this move last long enough to change the plan?
Language modelsOptional · off by default
Read unstructured text and draft explanations. Every number in AI-written text must match the recorded run exactly, or the text is rejected and the deterministic version is used. The AI layer is off unless configured.
What does this filing or memo actually say?
A morning's queue
Ranked by money at stake, not by alert count
The brain does not raise one alert per signal. It ranks proposals by money at stake, calibrated confidence and urgency, sends small items to an FYI list, and holds anything built on a stale or invalid feed.
Brain queue 6 items · illustrative
| Signal | Exposed | Money at stake(adverse case) | Sandbox check | Confidence | Outcome |
|---|---|---|---|---|---|
| Copper +1.6σ in five sessionsVersioned rule | 3 projects, one cable supplier | +$41M P80 capex | Holds | 0.81 | Act · CFO |
| Contractor's listed peer −27% in a monthVersioned rule | 2 units in construction | +$18M contingency | Holds | 0.64 | Review · projects |
| Margin within ½σ of a committee tripwireCommittee tripwire | 1 approved decision | Reopen | Holds | 0.72 | Revisit · committee |
| Freight index jumpVersioned rule | 2 projects | +$12M claimed | Engine finds +$2M | n/a | Flagged |
| Supplier filing: backlog figure changedSEC filing comparison | 1 project | +$3M | Holds | 0.55 | FYI |
| Rates feed stale for 3 daysFeed gate | n/a | n/a | Not run | n/a | Held |
Illustrative queue. Signals and figures are invented for the example and are not client data.
Corroboration counts
Independent sources reporting the same factor lift the confidence of a proposal. One noisy headline does not.
Disagreement is surfaced
When the engine's own re-run contradicts a proposal's claim, the proposal is flagged and ranked down instead of forwarded.
Bad data stops early
A signal whose source feed is stale, invalid or missing is held. It never becomes a proposal.
Guardrails
Automated attention, human authority
The brain is there to make sure nothing material goes unnoticed. It is not there to change your plan on its own.
Approval before any change
Every proposal is routed by your delegation-of-authority matrix, previewed before it is applied, and can be undone after.
The engine checks the claim
Each actionable proposal is re-run in a sandbox copy before it reaches anyone. The engine's figure replaces the detector's estimate.
Numbers checked in AI text
If a language model drafts the explanation, every number must match the recorded run exactly, or the deterministic text is used instead.
Learning proposes, people approve
Learning proposes new model versions. A person approves them before they become the default.
A kill switch
Operators can stop any automated job. A held job completes without changing anything.
Reasons are recorded
When someone declines a proposal they say why: a false alarm, sized too high or too low, already handled, or the wrong time. Those reasons feed the confidence of the next one.
Where it stands
What the brain does today, and what it does not yet
We label the brain Partial because part of its value depends on time.
- Runs today. The scheduled sequence from signal to approval queue, the sandbox check, the calibrated confidence, the kill switch and the record.
- Needs time. Outcome learning needs months of real decisions. Until then, confidence rests on the fitted models and the reasons people record when they act on a proposal.
- Data feeds. Market data comes from public sources such as FRED, SEC EDGAR and the ECB, for demonstration. Licensed market data is required before any customer use.
- Notifications. E-mail, Slack and Teams adapters never send without explicit configuration.
- Money. Capibud does not move funds or place orders. Approving a proposal changes a plan inside Capibud and nothing else.
See it on your own exposures. In a Decision Sprint the brain watches the factors your decision depends on, and the committee sets the tripwires it will check. How the committee sets them
Questions teams ask
Before the brain goes live
Every proposal it drafts lands in the decision record, and tripwires come from the capital committee.
Does the brain change the plan on its own?
No. It drafts a proposal, verifies it in a sandbox and routes it to the right person's queue through your delegation of authority. Nothing changes the plan until that person approves it; the approved change is previewed, applied and can be undone.
What does the brain watch?
Market prices, company filings (SEC 10-K, 10-Q and 8-K filings and XBRL company facts), project actuals, contractors and counterparties, committee tripwires, and capacity and calendars, each mapped to the projects and holdings exposed to that factor.
Does AI produce the numbers?
No. Versioned rules and fitted, frozen models flag and estimate; the simulation engine produces every number. Language models are optional and off by default, and any number they write must match the recorded run exactly or the text is rejected.
Is the brain available today?
Partly. The scheduled pipeline from signal to approval queue runs today. Learning from outcomes needs months of real decisions, so that part is still ahead.
Point the brain at your book.
A Decision Sprint runs one real decision on your own files in six weeks, with the brain watching the factors that decision depends on.