Tools
Three thinking tools.
The argument, applied to a decision you already make.
These are thinking tools, not compliance audits and not legal advice. Each one takes the Dispatches argument — that a consequential automated decision should leave a record someone can examine, reproduce, and contest — and puts it to work on one real decision. They surface gaps rather than grade you, and none of them certifies anything. There is no login and no backend; nothing you enter leaves the page.
A quick read on one decision
Self-assessment
Hold one consequential automated decision up against the four properties of an admissible record and get a plain-language read-out of where the gaps are. A few minutes, no preparation needed.
Open the self-assessmentDraft a record
Decision-record builder
Work through what an admissible record of one decision would actually contain — inputs, rules, the path to contest it — and draft it as a document you can keep. A structured way to find what is missing before someone else does.
Open the builderRate a system on the 0–4 maturity rubric
The Admissibility Scorecard
Score a whole system — not a single decision — across provenance, replay, standing, and contestability on a five-level maturity scale. Read the resulting profile to find the binding constraint, the one thin dimension worth fixing first.
Open the scorecardA note on scope. Most automated decisions are trivial, reversible, and never contested, and building full admissibility into them would be waste dressed as rigor. Spend the cost only on the decisions that land hard on a specific person — the ones where someone might reasonably demand to contest the outcome. These tools are meant to help you decide which decisions those are, not to argue that every decision needs a record.
— Dispatches · Summit Cognitive