Decisions have a supply chain: provenance for the statements your decisions rest on

Who this is for. Leaders and reviewers in organizations whose decisions have to be explained later to someone who was not there: program offices and coalition partners in Defense, compliance and audit functions in financial services, clinical governance committees in healthcare, boards and general counsel in commercial enterprises, and the chief information and data officers who are asked to make the trail possible.

1. A brief goes out on Thursday

Follow one document back to where it came from. The brief is four pages. Paragraph three carries a number: the delivery window, the dose, the limit, the price band, whatever the organization trades in. That number was settled on a Tuesday call between three people, and the call is the only place it was ever said out loud with the reasoning attached. Paragraph five carries a constraint. It was set in a planning meeting in June, and it was revised in September in a conversation that produced a short email and no change to the planning document. The recommendation in the final paragraph rests on a position one person argued, persuasively, in a chat thread that has since scrolled out of reach.

The brief is accurate. It is accurate because its author happened to be in every one of those rooms, remembered the September revision and knew which chat thread to search. It is defensible for exactly as long as those people are available to vouch for it. The day the author moves on, the brief becomes a document that says things without saying why, and the organization has to decide whether to trust it or redo the work.

2. The supply chain nobody tracks

Every decision rests on statements made somewhere. The statements have sources, dates and authors; they are made in meetings, recordings, minutes, chat and email; they supersede each other over time and they sometimes contradict each other. That chain of statements, from the first time a number was said to the moment it appeared in a signed document, is the supply chain of the decision, and in most organizations it is untracked.

It is untracked for an ordinary reason: the statements are perishable and scattered. A 2018 survey of a thousand employees at larger US organizations found that 42 percent of the knowledge needed to do a role was held only by the person in it, and that employees spend 5.3 hours a week waiting for knowledge a colleague holds or recreating it.¹ Microsoft's 2023 Work Trend Index, drawn from 31,000 workers in 31 countries, found that 56 percent of people find it hard to summarize what happened in a meeting, 55 percent say next steps at the end of a meeting are unclear, and people are in three times as many online meetings and calls as they were in early 2020.² More statements are being made, in more rooms, and fewer of them are landing anywhere.

The failure is visible at the moment someone asks why. A regulator asks why a limit was set where it was. A program office asks why a milestone moved. A coalition partner asks why a route was ruled out. A clinical governance committee asks why a protocol changed. In each case the organization produces a reconstruction: people are asked what they remember, emails are searched, the recording archive is opened and nobody has time to listen to forty minutes of it. The answer that comes back is sincere and unverifiable. It is a memory presented as a record.

3. Why reconstruction fails

Peter Drucker's test for whether a decision has actually been made is a list of names: the people who have to be told, the people who will be affected, the person accountable, and the deadline; he adds that a decision should be reviewed against its assumptions at a time agreed in advance, which is only possible if the assumptions were recorded.³ Most organizations fail the second half of that test not because they do not review decisions but because they cannot find what the decision rested on. The reasoning was never a record.

The records management profession has a precise word for what a reconstruction lacks. ISO 15489 asks that an authoritative record be reliable, which it defines as a full and accurate account created at the time of the event, or soon after, by people with direct knowledge of the facts; and that it have integrity, meaning it is complete and unaltered, and that any alteration can be identified along with the person who made it.⁴ A reconstruction assembled months later from memory fails both tests by construction. It was not created near the event, and nobody can say what in it has been altered by the passage of time.

The practical consequence is that the organization's most consequential statements, the ones decisions actually rest on, are the ones with the weakest evidentiary standing. The structured record is auditable to the row. The decision that produced the row is a story.

4. Provenance as a property of the statement

/TRUTH, Bluestaq's document-intelligence product, shipping today, tracks the supply chain by making provenance a property of the statement itself. Documents, meetings, transcripts, recordings and email go in. /TRUTH distills them into atomic claims: single statements of what a source says, each carrying its source, its date and how it was grounded. The Tuesday call becomes a claim about the number, pointing at the recording, the date and the passage. The September conversation becomes a claim about the constraint, pointing at the email. The chat thread becomes a claim about the position, pointing at the message.

Grounding matters as much as source. A claim records how it was derived from the source: whether the source says it explicitly, whether it was inferred, whether it was asserted by a named person. Sources can carry descriptors and a reliability grade, so a reviewer can see not just where a statement came from but how far to trust the place it came from. This is the ISO 15489 reliability test applied at the level of the sentence: created close to the event, by people with direct knowledge, with the derivation on record.

Verification is then a review pass an authorized user runs, not a property that appears by magic. The result is a record of sourced, verified statements that a decision can rest on, where every one of them can be followed back to the room it came from by someone who was never in it.

5. Supersession and conflict: a record that keeps its history

A supply chain changes. The June constraint was revised in September, and a record that only held the September version would have lost the reason the June one existed, while a record that held both without saying which was current would be worse than useless. /TRUTH handles this with supersession. When a newer source changes a position, the newer claim supersedes the older one, and the older stays in the record, marked superseded and kept. Source dates decide the order; an authorized editor can override. What changed, when, and what it replaced is a property of the record, not a memory.

Sometimes nothing supersedes anything: two sources simply disagree. The vendor report says the 14th; the contract modification says the 30th. Here /TRUTH refuses to flatten. The conflict is surfaced with both sides cited, and it stays surfaced until a source resolves it or an authorized person does. A brief built on a conflicted claim shows the conflict; a summary would have picked one and hidden the choice.

6. Briefs as views over claims

The last link in the chain is the document itself. In /TRUTH, rendered briefs, pages and FAQs are views over claims rather than copies of them. Every sentence in a rendered brief carries a sources line naming the claims it rests on; click the sentence and see the claim, the source and the date. Because the brief is a view, it stays current as the record changes: when the September revision supersedes the June constraint, the brief shows the September source and the June one it replaced, without anyone remembering to update it.

Watch lists close the loop for the people who depend on a statement. Put a claim on a watch list and a change digest tells you when it changes or now conflicts with something new. The author of Thursday's brief does not have to be in the room where the number moves; the record was, and it says so.

The same trail is available to software. Agents connect over MCP, with OAuth 2.1 and PKCE, and act as the user, so an agent drafting a brief sees only what its user may see and cites what it used. Systems connect over a REST integration API. Access is granted per project, so a restricted discussion and a program's working notes sit side by side under different permissions. Every answer, whether a person or an agent asked, cites its claims and sources.

7. What this gives the person who asks why

Now the brief on Thursday cites the Tuesday call. The revised constraint shows its September source and the June one it replaced. The recommendation points at the chat thread, dated. The author did not need to be in the room; the record was. And when a regulator, a program office, a coalition partner or a committee asks why, the answer is a trail, not a memory: a chain of sourced, verified statements that someone who was never present can follow to the end.

For a program office, that is a milestone whose every movement is on record. For a compliance function, it is a limit whose rationale can be produced on request with the date it was set and the date it was revised. For a clinical committee, it is a protocol whose current and previous versions are both in the record, marked. For a coalition partner, it is a constraint that can be traced to the call that set it. The organization keeps its ability to explain itself after the people who could explain it have moved on, and it keeps what it has built: every topic you can access exports whole, claims included, and /TRUTH runs as a dedicated instance for your organization.

Bluestaq runs its own company strategy on /TRUTH. It is one of four equal products, with /MARKET, BLUESTAQ /ARQ^™ and /LAUNCH, on one secure, open platform.