A governed simulation, not a standard AI answer.
Quorum is the platform your team analyses a high-stakes matter on. What comes out is an evidence base you can put in front of a board, a regulator or a counterparty. Bounded scope. Your named sign-off. A complete audit trail.
What Quorum is.
Quorum is a research production system with quality assurance built into it. A matter comes in; a decision-ready, sourced simulation comes out. The work is carried out by a system of independent perspectives chosen so they do not share blind spots. Their findings are cross-examined and stress-tested rather than averaged. Every claim is traced to a source and graded for the confidence it deserves. What could not be established is reported as an open question rather than papered over. Disagreement is not smoothed away. It is surfaced as a finding in its own right, because a question the field has not settled is important information for a decision-maker.
Most decisions are not one matter. They are competing paths.
A consequential decision rarely arrives with a single option on the table. There are three or four routes, each with a case behind it and an advocate in the room. The work is not to answer one question. It is to set the routes against each other on the same evidence, on equal terms, and establish which of them survives.
The paths are stated, not assumed
Each route is written as a proposition capable of failing, together with the conditions that would make it the right one. A route nobody can state that way is not yet an option. It is a preference.
Each is evidenced on its own merits
The same system, the same grading, the same declared absence, applied route by route. A well-evidenced option and a thin one cannot borrow each other's credibility, because the grades sit on the claims rather than on the document.
Then they are reconciled
Where routes rest on contradictory evidence, the contradiction is argued out under structure and ruled on by your reviewer. Discarded routes stay in the record with the reason they were set aside, so the decision can be defended against the option nobody chose.
What you hold at the end is never a recommendation without its basis: the recommendation comes with a graded comparison of the routes and the conditions under which each becomes the right one, and it does not tell you which path will succeed. It establishes what each one requires to be true.
Not a report you have to take on faith.
A Quorum simulation delivers six components. A short read for the decision-maker, a long read for the sceptic, and a record for whoever asks afterwards.
An executive priorities read
A decision-shaped summary: what the evidence supports, what it does not, where the risk sits, and which open questions would most change the picture. Stated up front rather than discovered in an appendix.
A graded evidence base
The substance of the simulation, organised by claim rather than by narrative. A reader moves from any statement in the executive read to the claim supporting it, and from that claim to the material behind it.
Per-claim provenance and confidence
Every claim carries its own evidence record: supporting sources, how far they corroborate one another, whether they are genuinely independent, and an explicit confidence grade. A simulation can carry a strongly evidenced core and a weakly evidenced periphery; the deliverable shows which is which.
Declared absence and open questions
A dedicated section stating what could not be established: coverage the system could not reach, claims that failed corroboration, conclusions the process declined to draw. Short, blunt, and deliberately prominent.
A full evidence report
The complete simulation, sub-question by sub-question, including where the system disagreed and how that was resolved, for the reader who wants to check the working rather than accept the summary.
An audit record
The pre-registered brief and every revision with its stated reason, the sub-question structure, the sources, and every human gate with who cleared it. Reconstructable long after the people involved have moved on.
A decision examined from every side that could break it.
Mandate definition
Before any research work begins, the matter goes through governance. It is framed independently by several parties, so the framing is not inherited from whoever raised it. A challenger then attacks it: the assumptions hidden inside it, the criteria that could never be falsified, the way it has been scoped to produce a particular answer. A brief that cannot fail is refused rather than run.
What survives is broken into discrete, answerable questions and signed off by your reviewer. The brief is then fixed. Scope is locked before any work begins, and it does not move without a record.
Revisions are allowed, since the world changes and so do questions. But each is versioned and leaves a permanent record with a stated reason, so nobody can quietly move the goalposts to match the result.
Evidence production and assurance
Each question is researched in parallel across the system. The system is deliberately diversified, because perspectives that share their assumptions will agree with each other reliably and tell you nothing new.
Findings are fused into single claims, then verified for provenance: not merely that a source exists, but that it supports the claim being made and is independent of the other sources behind it. Whoever gathered the evidence does not grade it. That separation of duties is what makes audit and editorial work trustworthy.
Where the system disagrees, the disagreement is escalated to structured challenge and ruled on by an accountable reviewer on your side. It is never averaged into a middle position that nobody would defend.
The result is rolled up into the executive read and the full evidence report.
Decision simulation
A consequential decision is rarely wrong because a fact was wrong. It is wrong because it was only ever examined from the angle of the people who wanted it to work. So the system does not only gather evidence for the matter as framed. It runs the same decision through the lenses most likely to break it, and reports what each one sees.
Second-order effects: what this sets in motion beyond the intended result
Who is harmed if it turns out wrong, and how visibly
What would have to be true for this to fail, stated as testable conditions
How it reads to a regulator examining it afterwards
How it looks now, against how it looks in three years
Where the lenses agree, that agreement is itself evidence. Where they diverge, the divergence is preserved as a finding and ruled on rather than averaged away, because the divergence is usually the thing worth knowing.
The four things that have to be true.
Independence: agreement only counts when it is independent
The system is constructed to disagree, not to converge. Independence between sources is established per claim rather than inferred from the fact that several sources say the same thing. Where they genuinely disagree, that is recorded as a finding.
Evidence: every claim carries its own provenance
Verification checks that a source supports the claim, not merely that it exists. Corroboration level and confidence are attached per claim. Claims that cannot be corroborated become open questions.
Governance: scope is locked before work begins, a human owns the verdict
The pre-registered brief, the adversarial intake, and defined sign-off gates held by your own people. The platform does not release unreviewed work.
Accountability: the record outlives the decision
The complete record: the matter, revisions, sources, approvals. Bidirectional traceability from any conclusion back to its evidence and forward from the pre-registered brief.
If you are weighing one of these, you are in the right place.
Should we enter this market, and what would have to be true for that to be wrong?
Is this technology claim real, or has it been announced and demonstrated but never shipped?
Who is actually ahead in this field, on the evidence rather than the narrative?
Does this interpretation hold up, and what is the strongest argument against it?
What is the disconfirming evidence for the thesis we already believe?
Is this supplier, partner or target what it presents itself as?
What does the primary literature actually support, as distinct from what is widely repeated?
Where is our position weakest, and who is most likely to attack it?
The common shape: the matter is consequential enough that being confidently wrong is expensive, and the answer will be examined by someone who was not in the room.
A market-entry matter.
The matter as it arrived: “Confirm that the regulatory environment in this market supports our entry.”
What adversarial intake did to it: the brief was refused in that form. As written it could only produce confirmation. There was no result that would have counted as a negative answer. It was reframed as two falsifiable questions: what does the current regime require of a market entrant in this category, and what has changed or is under consultation that would alter those requirements within the planning horizon.
What the simulation produced: a graded evidence base against both questions, with requirements traced to primary regulatory instruments rather than to secondary commentary; two claims widely repeated in trade press that could not be corroborated against any primary source, flagged as such; one area of genuine disagreement inside the system, escalated and ruled on with the dissenting position recorded; and an explicit statement of what could not be established, namely the timing of one pending consultation.
What the commissioning team can do with that: proceed, with the pending consultation carried as a tracked condition rather than as an unexamined assumption.
Why this example is here: the value was not the answer. It was that the brief as originally written could only have produced a comfortable yes.
What Quorum does not do.
It does not make the decision. It produces the evidence base a decision-maker acts on, and records the person on your side who signed it off.
It does not manufacture certainty. Where the evidence is thin, it says so, and “we could not establish this” is a result, not a gap to be papered over.
It does not replace your judgement, your counsel, or your regulator. It gives all three something they can examine.
It does not predict. It establishes what is currently supportable and how strongly.
Two platforms, each complete on its own.
Quorum answers one bounded matter to a standard you can act on with confidence. Compass maintains a standing picture: what is emerging in your field, where you stand, and what your own portfolio contains.
Neither is a prerequisite for the other. They share an evidence layer, so each makes the other more valuable, but Quorum is a complete product for an organisation that never licenses Compass.
Explore CompassStart with the decision that matters most this year.
If you are assessing this for procurement, security or legal review, the sign-off gates, data isolation, residency and retention are set out on Trust, governance and data.