Policy is made on evidence that is contested, incomplete, or misrepresented — and the consequences of getting it wrong are public. Augle runs the adversarial review that a well-resourced opposition will run anyway, before the decision is made rather than after.
A regulatory impact assessment that looks solid under friendly review will look very different under opposition scrutiny. The discount rate in a 30-year infrastructure model. The replication status of the behavioural study your intervention depends on. The jurisdiction scope of the precedent your legal team cited. These things get found — the question is whether you find them first.
Questions policy teams run on Augle
The Methodologist evaluates construct validity across the entire evidence base — whether the methodology actually measures what the policy claims it measures. Standard CBA frameworks systematically undervalue tail risk over long time horizons. Augle surfaces that before the Treasury Select Committee does.
Many policy interventions are based on behavioural research conducted under controlled conditions with non-representative populations. The Contrarian identifies the generalisability objection — whether the effect size holds at population scale and in the real-world context the policy targets.
The Guardian's SVS validates statutory citations against current version and jurisdiction scope. Overruled decisions, superseded statutes, and cross-jurisdiction misapplications are flagged at Moderate severity — automatically, before they enter a briefing or a bill.
The Contrarian runs at maximum temperature to surface the adversarial framing an opposition researcher, a critical journalist, or a hostile expert witness will deploy. Unresolved Strong objections appear verbatim in the output — not softened, not summarised.
Each scenario below illustrates realistic deliberation behaviour across the policy research and legislative lifecycle.
“Does the proposed infrastructure bill's cost-benefit methodology adequately account for climate-adjusted risk over a 30-year horizon?”
The team knew the discount rate was a potential weakness. They didn't know the committee had already briefed a Stern Review expert. They found that out from Augle — before the session, not during it.
“Is the behavioural evidence base for the proposed sugar levy sufficient to support the projected public health outcomes at national scale?”
The distributional objection wasn't in the RIA. It would have been in the Health Select Committee's written evidence within 48 hours of publication. Three weeks of additional modelling prevented a headline they couldn't walk back.
“What does the evidence actually say about the relationship between stop-and-search powers and serious violent crime reduction — and what are the strongest objections to the published findings?”
A briefing that conflates localised deterrence with net crime reduction is a briefing that gets corrected in committee — publicly, and by the opposition researcher who already knows the distinction. It got fixed before it was published.
Submit bills, regulatory impact assessments, policy briefs, research papers, and statutory references. The Guardian's SVS validates citations, checks statutory versions, and confirms jurisdiction scope before any agent receives the evidence.
The Cartographer classifies every evidence node as Settled, Contested, or Unknown. The Methodologist assesses construct validity — whether the methodology measures what the policy claims. Confidence bounds are set as hard constraints before deliberation begins.
The Contrarian runs at maximum temperature to surface the objections an opposition researcher, expert witness, or critical journalist would raise. Every objection specifies a resolution condition. Unresolved Strong objections cannot be softened.
The Synthesizer produces a calibrated finding anchored to the evidence nodes registry — not the discourse thread. Confidence grades, unresolved objections verbatim, and reopen conditions. The full audit trail is exportable for regulatory compliance or FOI purposes.
The Contrarian is designed to surface the adversarial framing that a well-resourced opposition team will deploy. It steelmans every claim before challenging it — which means it finds the strongest version of the objection, not a weak one you can dismiss. If the objection is strong and unresolved, it appears verbatim in your output.
The Guardian's SVS checks retraction status, statutory version, and jurisdiction scope on every citation. A superseded regulation cited in a briefing, or a retracted study cited in an RIA, is flagged automatically — before it becomes a committee embarrassment or a FOI liability.
Every session produces a full audit trail — SVS verification outcomes, confidence downgrades, objections raised and resolved, and phase boundary evaluations. For regulated industries and parliamentary accountability, this is the record that demonstrates due diligence on the evidence base that informed a decision.
Join waitlist and run a session on the evidence behind your next policy decision.